PlexSheep commited on
Commit
0ecaf52
Β·
verified Β·
1 Parent(s): b58dad3

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +93 -1
README.md CHANGED
@@ -6,4 +6,96 @@ tags:
6
  pretty_name: ' Jigglypuff detection data'
7
  size_categories:
8
  - 10K<n<100K
9
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6
  pretty_name: ' Jigglypuff detection data'
7
  size_categories:
8
  - 10K<n<100K
9
+ ---
10
+
11
+ # Jigglypuff Detection Dataset
12
+
13
+ This dataset is designed for training machine learning models to detect and distinguish Jigglypuff (a Pokemon) from Kirby and other similar characters.
14
+
15
+ I mainly use this with my [Pummeluff-Detektor](https://github.com/PlexSheep/pummeluff_detektor)
16
+
17
+ ## Dataset Summary
18
+
19
+ The dataset contains approximately 5.5GB of images across three categories:
20
+ - Jigglypuff (Pokemon character)
21
+ - Kirby (Nintendo character)
22
+ - Other (Various other characters and images)
23
+
24
+ This dataset was specifically curated to solve the age-old problem of distinguishing between these two famously similar pink characters, along with detecting potential imposters.
25
+
26
+ ## Supported Tasks
27
+
28
+ - Image Classification
29
+ - Character Recognition
30
+ - Pink Blob Detection
31
+
32
+ ## Dataset Structure
33
+
34
+ ```
35
+ dataset/
36
+ β”œβ”€β”€ jigglypuff/ (~730MB)
37
+ β”œβ”€β”€ kirby/ (~4MB)
38
+ └── other/ (~4.8GB)
39
+ β”œβ”€β”€ other/
40
+ └── other2/
41
+ ```
42
+
43
+ ### Data Fields
44
+
45
+ The dataset is organized in a directory structure where each subdirectory represents a class label:
46
+ - `jigglypuff`: Contains Jigglypuff images
47
+ - `kirby`: Contains Kirby images, relevant because kirby and jigglypuff look similar
48
+ - `other`: Contains images of other characters and objects for negative examples
49
+
50
+ ### Data Splits
51
+
52
+ This is a raw image dataset without predefined splits. Users should implement their own train/test splits.
53
+
54
+ ## Dataset Creation
55
+
56
+ ### Source Data
57
+
58
+ Images were collected from two primary sources:
59
+ - [CS63 Pokemon Dataset](https://www.kaggle.com/datasets/alinapalacios/cs63-pokemon-dataset)
60
+ - [Nintendo Kirby Dataset](https://www.kaggle.com/datasets/turbihao/nintendo-kirby-dataset)
61
+
62
+ As of writing this, no additional images were added.
63
+
64
+ ### Annotations
65
+
66
+ The dataset uses directory names as implicit labels. No additional annotation files are provided.
67
+
68
+ ## Considerations for Using the Dataset
69
+
70
+ ### Social Impact of Dataset
71
+
72
+ This dataset was created for educational and entertainment purposes. It demonstrates practical machine learning concepts through a fun and accessible example.
73
+
74
+ ### Discussion of Biases
75
+
76
+ The dataset may contain biases in terms of:
77
+ - Art styles (anime/game artwork vs. fan art)
78
+ - Image quality variations
79
+ - Pose distributions
80
+ - Background variations
81
+ - Only images of Pokemon or Kirby, no unrelated images
82
+
83
+ ### Other Known Limitations
84
+
85
+ - Limited to specific art styles
86
+ - May not generalize well to real-world pink objects
87
+ - Focused on frontal/profile views
88
+
89
+ ## Additional Information
90
+
91
+ ### Dataset Curators
92
+
93
+ Dataset curated by @plexsheep
94
+
95
+ ### Contributions
96
+
97
+ Feel free to open a discussion on the dataset repository page if you have:
98
+ - Suggestions for additional images
99
+ - Quality improvements
100
+ - Documentation updates
101
+ - Use case examples