scfive commited on
Commit
6c17d81
·
verified ·
1 Parent(s): 0638912

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +57 -15
README.md CHANGED
@@ -1,15 +1,57 @@
1
- typing-extensions>=4.5.0
2
- ultralytics==8.0.196
3
- opencv-python==4.8.1.78
4
- numpy==1.24.3
5
- torch==2.1.0
6
- torchvision==0.16.0
7
- transformers==4.30.2
8
- Pillow==10.0.0
9
- networkx==3.1
10
- openai==1.3.0
11
- torch-geometric==2.3.1
12
- protobuf==3.20.3
13
- tensorboard==2.13.0
14
- gradio==4.0.0
15
- supervision==0.3.0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ title: Glad8tr Video Analysis
3
+ emoji: 🎥
4
+ colorFrom: blue
5
+ colorTo: indigo
6
+ sdk: gradio
7
+ sdk_version: 4.0.0
8
+ app_file: app.py
9
+ pinned: false
10
+ ---
11
+
12
+ # Glad8tr Video Analysis
13
+
14
+ This is a video analysis application that uses computer vision to detect objects, analyze poses, and provide cognitive state analysis. The application is deployed on Hugging Face Spaces.
15
+
16
+ ## Features
17
+
18
+ - Object detection and tracking
19
+ - Pose estimation and analysis
20
+ - Scene context analysis
21
+ - Cognitive state analysis
22
+ - Real-time video processing
23
+
24
+ ## Usage
25
+
26
+ 1. Upload a video file (supported formats: MP4, AVI, MOV)
27
+ 2. The application will process the video and provide:
28
+ - Object detection results
29
+ - Pose analysis
30
+ - Scene context information
31
+ - Cognitive state analysis
32
+ 3. Download the processed video with annotations
33
+
34
+ ## Technical Details
35
+
36
+ - Built with PyTorch and YOLOv8
37
+ - Uses Gradio for the web interface
38
+ - Optimized for Hugging Face Spaces deployment
39
+ - Processes videos in real-time with frame sampling
40
+
41
+ ## Limitations
42
+
43
+ - Processing is limited to the first 100 frames for demo purposes
44
+ - Maximum video resolution: 1920x1080
45
+ - Processing time depends on video length and complexity
46
+
47
+ ## Model Information
48
+
49
+ The application uses two YOLOv8 models:
50
+ 1. Object detection model (glad8trv8s.pt)
51
+ 2. Pose estimation model (glad8trv8s-pose.pt)
52
+
53
+ If the custom models are not available, the application will fall back to the base YOLOv8 models.
54
+
55
+ ## License
56
+
57
+ This project is licensed under the MIT License.