Dataset Viewer
Auto-converted to Parquet
id
int64
0
287
scene_name
stringlengths
8
12
dataset
stringclasses
3 values
video_suffix
stringclasses
2 values
question
stringlengths
794
990
0
scene0251_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 2 chair, 3 window, 5 trash bin, 3 door, 2 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
1
scene0277_02
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 bed, 1 chair, 1 window, 1 radiator, 2 trash bin, 1 door, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
2
scene0304_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 backpack, 1 trash bin, 1 door in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
3
scene0307_02
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 chair, 3 window, 1 refrigerator, 1 washing machine, 1 radiator, 5 door, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
4
scene0316_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 2 chair, 1 door, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
5
scene0328_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 window, 1 counter, 1 refrigerator, 1 microwave, 1 oven in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
6
scene0338_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 counter, 2 trash bin, 1 door in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
7
scene0353_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 bed, 3 chair, 1 sofa, 1 window, 1 bookshelf, 1 refrigerator, 4 pillow, 2 backpack, 1 closet, 1 trash bin, 1 door, 2 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
8
scene0354_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 8 chair, 1 window, 1 tv, 1 trash bin, 1 door, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
9
scene0356_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 bed, 1 chair, 1 window, 1 radiator, 1 telephone, 2 trash bin, 3 door, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
10
scene0357_01
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 window, 1 towel, 1 backpack, 2 door, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
11
scene0378_01
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 4 chair, 1 window, 2 bookshelf, 1 monitor, 1 printer, 1 clock, 1 trash bin, 1 door, 3 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
12
scene0412_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 6 trash bin, 2 door, 5 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
13
scene0426_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 bed, 3 chair, 1 window, 3 pillow, 4 monitor, 1 lamp, 1 tv, 1 backpack, 1 door, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
14
scene0616_01
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 7 chair, 1 window, 1 lamp, 1 radiator, 2 trash bin, 2 door, 2 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
15
scene0621_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 27 chair, 1 door, 5 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
16
scene0629_01
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 bed, 1 chair, 1 plant, 1 mirror, 2 door, 2 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
17
scene0633_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 chair, 1 window, 1 pillow, 1 lamp, 1 clock, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
18
scene0643_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 5 chair, 1 window, 1 bookshelf, 1 refrigerator, 2 monitor, 1 keyboard, 1 telephone, 2 trash bin, 1 door, 5 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
19
scene0645_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 2 bed, 3 chair, 1 sofa, 2 window, 14 pillow, 3 lamp, 1 towel, 1 tv, 1 nightstand, 3 backpack, 2 mirror, 1 telephone, 3 trash bin, 3 door, 3 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
20
scene0647_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 chair, 2 sofa, 1 tv, 1 backpack, 2 door, 3 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
21
scene0648_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 2 bed, 2 chair, 1 window, 1 bookshelf, 5 pillow, 1 lamp, 2 plant, 2 backpack, 1 closet, 1 fan, 1 door, 2 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
22
scene0651_02
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 4 chair, 1 sofa, 1 counter, 1 refrigerator, 1 trash bin, 2 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
23
scene0653_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 8 chair, 2 window, 7 monitor, 2 keyboard, 2 backpack, 1 clock, 2 trash bin, 1 door, 6 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
24
scene0658_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 bed, 1 chair, 1 window, 1 lamp, 1 trash bin, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
25
scene0663_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 3 chair, 1 window, 3 monitor, 2 keyboard, 3 backpack, 2 telephone, 2 trash bin, 1 door, 2 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
26
scene0664_02
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 counter, 2 towel, 1 mirror, 1 trash bin, 1 door in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
27
scene0665_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 18 chair, 1 door, 9 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
28
scene0671_01
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 4 chair, 2 pillow, 1 plant, 2 backpack, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
29
scene0684_01
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 chair, 2 window, 1 telephone, 1 door, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
30
scene0685_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 2 trash bin, 1 door, 2 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
31
scene0686_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 mirror, 1 trash bin in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
32
scene0693_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 window, 1 towel, 3 mirror, 1 trash bin, 1 door in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
33
scene0695_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 bed, 2 chair, 1 window, 1 bookshelf, 1 pillow, 1 monitor, 1 lamp, 1 backpack, 1 trash bin, 1 door, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
34
scene0696_01
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 mattress, 10 chair, 1 sofa, 2 pillow, 2 tv, 1 radiator, 1 door, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
35
scene0697_01
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 2 bed, 2 window, 3 pillow, 1 lamp, 1 tv, 2 nightstand, 1 closet, 1 mirror, 1 trash bin in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
36
scene0699_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 bed, 1 chair, 2 window, 2 pillow, 1 nightstand, 2 backpack, 1 closet, 2 door, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
37
scene0700_02
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 4 chair, 1 window, 2 monitor, 1 keyboard, 2 backpack, 1 radiator, 1 telephone, 2 trash bin, 1 door, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
38
scene0701_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 2 sofa, 1 backpack, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
39
scene0702_02
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 telephone, 1 trash bin, 1 door in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
40
scene0704_01
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 counter, 1 printer, 4 trash bin, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
41
scene0435_02
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 2 bed, 2 chair, 1 refrigerator, 8 pillow, 2 lamp, 1 tv, 1 nightstand, 1 backpack, 1 trash bin, 1 door, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
42
scene0441_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 mirror, 3 trash bin, 1 door in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
43
scene0458_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 window, 1 towel, 1 mirror, 2 trash bin, 1 door in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
44
scene0461_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 2 chair, 1 window, 2 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
45
scene0462_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 refrigerator, 1 printer, 2 trash bin, 2 door in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
46
scene0474_04
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 3 chair, 1 sofa, 2 window, 1 bookshelf, 3 pillow, 2 monitor, 1 backpack, 1 trash bin, 1 door, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
47
scene0488_01
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 2 counter, 2 refrigerator, 1 microwave in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
48
scene0494_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 4 chair, 1 window, 1 trash bin, 1 door, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
49
scene0496_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 16 chair, 4 window, 2 trash bin, 1 door, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
50
scene0500_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 38 chair, 7 window, 1 door, 2 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
51
scene0518_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 2 chair, 1 sofa, 2 pillow, 1 tv, 1 backpack, 2 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
52
scene0550_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 window, 1 mirror, 1 radiator, 1 trash bin, 1 door in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
53
scene0553_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 backpack, 1 mirror, 1 trash bin, 1 door in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
54
scene0565_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 4 chair, 3 window, 4 monitor, 2 printer, 1 keyboard, 1 trash bin, 2 door, 2 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
55
scene0568_02
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 4 chair, 1 sofa, 1 bookshelf, 1 tv, 1 radiator, 2 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
56
scene0574_01
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 4 towel, 1 mirror, 1 trash bin, 2 door in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
57
scene0575_01
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 10 chair, 1 window, 2 tv, 1 trash bin, 1 door, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
58
scene0580_01
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 bed, 1 chair, 1 window, 1 bookshelf, 2 pillow, 1 lamp, 1 nightstand, 2 backpack, 1 mirror, 1 trash bin, 2 door, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
59
scene0591_01
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 2 chair, 1 sofa, 1 window, 2 pillow, 4 monitor, 2 plant, 2 keyboard, 1 backpack, 1 telephone, 2 trash bin, 1 computer mouse, 1 door, 2 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
60
scene0593_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 sofa, 2 window, 1 printer, 1 backpack, 1 fan, 2 trash bin, 1 door, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
61
scene0598_02
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 2 bookshelf in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
62
scene0608_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 3 chair, 1 sofa, 1 window, 2 pillow, 1 lamp, 1 tv, 3 plant, 1 clock, 1 guitar, 1 door, 3 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
63
scene0025_01
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 2 chair, 2 sofa, 3 window, 1 pillow, 4 monitor, 1 keyboard, 1 fan, 1 telephone, 1 door, 2 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
64
scene0030_02
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 17 chair, 3 window, 6 bookshelf, 1 clock, 1 trash bin, 1 door, 6 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
65
scene0046_01
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 bed, 1 chair, 1 lamp, 1 towel, 1 tv, 1 nightstand, 2 backpack, 1 mirror, 1 trash bin, 2 door, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
66
scene0050_01
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 2 chair, 1 sofa, 1 lamp, 1 printer, 2 backpack, 1 piano, 4 door, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
67
scene0077_01
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 window, 1 counter, 3 printer, 2 trash bin, 2 door, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
68
scene0081_02
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 27 chair, 7 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
69
scene0086_02
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 window, 1 plant, 1 mirror, 1 trash bin, 3 door in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
70
scene0088_01
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 18 chair, 2 window, 1 trash bin, 1 door, 2 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
71
scene0095_01
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 17 chair, 1 window, 10 monitor, 1 radiator, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
72
scene0100_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 window, 1 mirror, 1 trash bin, 1 door in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
73
scene0144_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 bed, 3 chair, 1 window, 1 pillow, 1 lamp, 1 printer, 1 nightstand, 1 door, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
74
scene0146_01
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 backpack, 1 mirror, 1 trash bin, 1 door in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
75
scene0149_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 chair, 1 window, 2 counter, 1 refrigerator, 1 microwave, 1 trash bin, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
76
scene0153_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 3 towel, 1 mirror, 1 door in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
77
scene0164_02
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 counter, 1 refrigerator, 1 towel, 1 backpack, 1 microwave, 1 door in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
78
scene0169_01
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 22 chair, 2 window, 1 backpack, 1 trash bin, 1 door, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
79
scene0193_01
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 bed, 1 chair, 1 window, 1 door, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
80
scene0196_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 23 chair, 1 backpack, 1 trash bin, 1 door, 5 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
81
scene0203_01
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 2 chair, 2 sofa, 2 window, 1 bookshelf, 4 pillow, 2 door, 3 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
82
scene0207_02
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 bed, 1 chair, 1 sofa, 2 window, 1 counter, 1 refrigerator, 1 pillow, 1 monitor, 1 printer, 2 towel, 1 nightstand, 2 backpack, 1 closet, 1 microwave, 1 mirror, 1 radiator, 1 trash bin, 2 door, 2 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
83
scene0208_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 2 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
84
scene0221_01
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 2 bed, 3 chair, 5 pillow, 2 lamp, 1 nightstand, 1 microwave, 1 radiator, 1 trash bin, 2 door, 2 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
85
scene0222_01
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 2 bed, 2 chair, 2 window, 2 refrigerator, 3 pillow, 1 lamp, 1 nightstand, 1 fan, 1 radiator, 3 trash bin, 2 door, 2 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
86
scene0231_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 6 chair, 1 sofa, 3 window, 1 counter, 1 refrigerator, 1 pillow, 2 lamp, 1 plant, 1 backpack, 1 microwave, 1 clock, 2 trash bin, 1 door, 2 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
87
scene0246_00
scannet
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 bed, 1 chair, 6 pillow, 2 lamp, 1 towel, 1 nightstand, 2 mirror, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
88
7b6477cb95
scannetpp
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 4 ceiling light, 2 whiteboard, 5 monitor, 2 heater, 1 telephone, 1 power strip, 4 chair, 3 table, 1 trash can, 1 door, 1 bookshelf, 2 keyboard, 1 cup, 1 computer mouse in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
89
c50d2d1d42
scannetpp
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 2 trash can, 2 heater, 3 computer tower, 1 telephone, 6 monitor, 1 headphones, 1 door, 4 ceiling light, 4 chair, 1 whiteboard, 2 keyboard, 2 computer mouse, 1 coat hanger in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
90
cc5237fd77
scannetpp
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 3 monitor, 2 door, 2 chair, 2 bookshelf, 2 computer tower, 1 suitcase, 1 laptop, 1 ceiling light, 1 keyboard in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
91
acd95847c5
scannetpp
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 2 computer tower, 4 chair, 1 door, 4 ceiling light, 2 trash can, 1 bookshelf, 2 power strip, 1 whiteboard, 3 table, 2 keyboard, 2 computer mouse, 3 monitor, 2 telephone, 1 laptop, 2 heater in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
92
fb5a96b1a2
scannetpp
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 12 ceiling light, 1 sofa, 18 chair, 2 table, 1 bookshelf, 3 whiteboard, 1 heater, 1 power strip, 1 telephone, 1 picture, 1 computer mouse in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
93
a24f64f7fb
scannetpp
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 monitor, 1 chair, 1 door, 1 picture, 2 table, 1 ceiling light, 1 keyboard, 1 computer mouse in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
94
1ada7a0617
scannetpp
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 2 bucket, 2 whiteboard, 4 chair, 1 heater, 3 keyboard, 4 monitor, 1 computer tower, 2 telephone, 4 computer mouse, 1 door in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
95
5eb31827b7
scannetpp
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 3 whiteboard, 2 table, 1 bookshelf, 3 picture, 6 monitor, 2 computer mouse, 1 computer tower, 3 trash can, 1 door, 2 laptop, 1 keyboard, 1 heater, 1 telephone in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
96
3e8bba0176
scannetpp
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 1 coat hanger, 1 paper bag, 5 chair, 2 whiteboard, 1 door, 2 trash can, 4 monitor, 2 keyboard, 3 table, 4 ceiling light in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
97
3f15a9266d
scannetpp
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 4 chair, 1 door, 2 bookshelf, 5 picture, 2 ceiling light, 1 exhaust fan, 3 cup, 1 telephone, 1 computer mouse, 1 keyboard, 2 monitor in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
98
21d970d8de
scannetpp
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 3 whiteboard, 40 chair, 1 tv, 1 telephone, 1 door, 12 ceiling light in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
99
5748ce6f01
scannetpp
.mp4
[Task] This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix. [Rule] 1. We provide the cared categories and their corresponding instances: 2 whiteboard, 2 ceiling light, 1 cup, 5 chair, 2 door, 1 table in this scene. You should ONLY consider these categories. 2. You should use 0 to represent an empty grid and different numbers to represent different instances. 3. If there are multiple instances in a category, please draw all of them. 4. If multiple objects overlap in the same location, keep only the one with the largest area. [Output] Your task is to fill in the locations of ALL objects (instance level). Please respond in the following JSON format: {"grid": 10x10 matrix in list of list, "legend": {0: "empty", 1: "object 1", 2: "object 2", ....}}
End of preview. Expand in Data Studio

No dataset card yet

Downloads last month
3