KRadim commited on
Commit
56dc528
·
verified ·
1 Parent(s): 7666d97

Update README.md

Browse files

Add README.md file.

Files changed (1) hide show
  1. README.md +141 -0
README.md CHANGED
@@ -127,3 +127,144 @@ pretty_name: AwA-Pose-Lite
127
  size_categories:
128
  - 10K<n<100K
129
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
127
  size_categories:
128
  - 10K<n<100K
129
  ---
130
+
131
+ ## **Dataset Summary**
132
+
133
+ AwA Pose, a quadrupedal keypoint detection dataset that offers richer annotations and greater species diversity than existing datasets. The data supports efficiency, advances in research on generalized keypoint detection in animals.
134
+
135
+ The dataset was explored, filtered, and stored in a [pyarrow type](https://arrow.apache.org/docs/python/index.html). See : [AwA2_dataset_analysis](https://www.kaggle.com/code/radimkzl/awa2-dataset-analysis)
136
+
137
+ ***Original Kaggle dataset:*** [AwA2_dataset](https://www.kaggle.com/datasets/radimkzl/awa2-dataset)
138
+
139
+ The original data comes from two sources and is described in:
140
+
141
+ - [A Novel Dataset for Keypoint Detection of
142
+ Quadruped Animals from Images](https://arxiv.org/pdf/2108.13958)
143
+ - [GitHub: prinik/AwA-Pose](https://github.com/prinik/AwA-Pose/tree/main)
144
+ - [Animals with Attributes 2](https://cvml.ista.ac.at/AwA2/)
145
+
146
+ *The dataset contains:*
147
+ - Lite version of keypoints dataset
148
+ - Full version of keypoints dataset
149
+
150
+ *All version contains:*
151
+ - train dataset: 90%
152
+ - validation dataset: 5%
153
+ - test dataset: 5%
154
+
155
+ ## **Description of data in the dataset**
156
+
157
+ *Class names:*
158
+ - antelope
159
+ - bobcat
160
+ - buffalo
161
+ - chihuahua
162
+ - collie
163
+ - cow
164
+ - dalmatian
165
+ - deer
166
+ - elephant
167
+ - fox
168
+ - german+shepherd
169
+ - giant+panda
170
+ - giraffe
171
+ - grizzly+bear
172
+ - hippopotamus
173
+ - horse
174
+ - leopard
175
+ - lion
176
+ - moose
177
+ - otter
178
+ - ox
179
+ - persian+cat
180
+ - pig
181
+ - polar+bear
182
+ - rabbit
183
+ - raccoon
184
+ - rhinoceros
185
+ - sheep
186
+ - siamese+cat
187
+ - squirrel
188
+ - tiger
189
+ - weasel
190
+ - wolf
191
+ - zebra
192
+
193
+ ## **Columns description**
194
+
195
+ | name column| description of column |
196
+ |:--------------|:--------------------------------|
197
+ | id | id number of records |
198
+ | right_eye | keypoint values [x,y] |
199
+ | right_earbase | keypoint values [x,y] |
200
+ | right_earend | keypoint values [x,y] |
201
+ | right_antler_base | keypoint values [x,y] |
202
+ | right_antler_end | keypoint values [x,y] |
203
+ | left_antler_base | keypoint values [x,y] |
204
+ | left_antler_end | keypoint values [x,y] |
205
+ | left_earbase | keypoint values [x,y] |
206
+ | left_earend | keypoint values [x,y] |
207
+ | left_eye | keypoint values [x,y] |
208
+ | nose | keypoint values [x,y] |
209
+ | upper_jaw | keypoint values [x,y] |
210
+ | lower_jaw | keypoint values [x,y] |
211
+ | mouth_end_right | keypoint values [x,y] |
212
+ | throat_base | keypoint values [x,y] |
213
+ | neck_base | keypoint values [x,y] |
214
+ | neck_end | keypoint values [x,y] |
215
+ | back_base | keypoint values [x,y] |
216
+ | back_middle | keypoint values [x,y] |
217
+ | back_end | keypoint values [x,y] |
218
+ | tail_base | keypoint values [x,y] |
219
+ | body_middle_right | keypoint values [x,y] |
220
+ | bbox | bounding box dimension [x1, y1, x2, y2] |
221
+ | mouth_end_left | keypoint values [x,y] |
222
+ | throat_end | keypoint values [x,y] |
223
+ | tail_end | keypoint values [x,y] |
224
+ | front_left_thai | keypoint values [x,y] |
225
+ | front_left_knee | keypoint values [x,y] |
226
+ | front_left_paw | keypoint values [x,y] |
227
+ | front_right_thai | keypoint values [x,y] |
228
+ | front_right_paw | keypoint values [x,y] |
229
+ | front_right_knee | keypoint values [x,y] |
230
+ | back_left_knee | keypoint values [x,y] |
231
+ | back_left_paw | keypoint values [x,y] |
232
+ | back_left_thai | keypoint values [x,y] |
233
+ | back_right_thai | keypoint values [x,y] |
234
+ | back_right_paw | keypoint values [x,y] |
235
+ | back_right_knee | keypoint values [x,y] |
236
+ | belly_bottom | keypoint values [x,y] |
237
+ | body_middle_left | keypoint values [x,y] |
238
+ | name_file | name of file as string |
239
+ | name_class | name of class as string |
240
+ | image_base64s | image as string Base64 |
241
+ | image_width | value of with of image |
242
+ | image_height | value of height of image |
243
+ | image_license | text of licence of image |
244
+
245
+ ## **Notes on data**
246
+
247
+ &gt; If a *keypoints* contains `[-1.0, -1.0]`, it means that the point is not visible in the image. These points must be masked when training the model.
248
+
249
+ &gt; Images are stored as a Base64 string. They can be transformed using the function:
250
+
251
+ ```python
252
+ import base64
253
+ import io
254
+ from PIL import Image
255
+
256
+ def base64_to_img(base64_str):
257
+ img_bytes = base64.b64decode(base64_str)
258
+ img_buffer = io.BytesIO(img_bytes)
259
+ img = Image.open(img_buffer)
260
+
261
+ return img
262
+ ```
263
+
264
+ Find out more in [AwA2_dataset_analysis](https://www.kaggle.com/code/radimkzl/awa2-dataset-analysis#Create-AwA2-dataset-for-Hugging-Face...)
265
+
266
+ ## **Licensing Information**
267
+
268
+ Data for keypoints is licensed according to [GitHub: prinik/AwA-Pose](https://github.com/prinik/AwA-Pose/tree/main), this license is [MIT](https://github.com/prinik/AwA-Pose/tree/main?tab=MIT-1-ov-file#readme).
269
+
270
+ The license for the images is according to [Animals with Attributes 2](https://cvml.ista.ac.at/AwA2/), see data column `image_license `.