Golden Monkey Pose Estimation Dataset (Test Set)

# Golden Monkey Pose Estimation Dataset (Test Set) This dataset contains annotated images of Golden Monkey with 17 keypoints for pose estimation. This is the **test set**, intended for evaluating model performance in a standardized manner. ## Dataset Overview - **Species**: Golden Monkey - **Number of images**: 751 - **Number of annotations**: One annotation per image (single subject per image) - **Keypoint definition**: 17 keypoints (see below) - **Image format**: JPG - **Annotation format**: COCO-style JSON (compatible with common pose estimation frameworks) - **Purpose**: Model evaluation and benchmarking ## File Structure golden_monkey_test/ ├── imagesTest/ # All test images │ ├── 000000062461.jpg │ ├── 000000062462.jpg │ └── ... ├── annotations/ │ └── keypoints_test.json # Test annotation file └── README.md # This file ## Keypoint Definitions The 17 keypoints are defined in the following order (COCO keypoint convention): | Index | Keypoint | |-------|------------------| | 1 | nose | | 2 | left_eye | | 3 | right_eye | | 4 | left_ear | | 5 | right_ear | | 6 | left_shoulder | | 7 | right_shoulder | | 8 | left_elbow | | 9 | right_elbow | | 10 | left_wrist | | 11 | right_wrist | | 12 | left_hip | | 13 | right_hip | | 14 | left_knee | | 15 | right_knee | | 16 | left_ankle | | 17 | right_ankle | ## Annotation Format The annotation file (`keypoints_test.json`) follows the **COCO keypoint format** with the following top-level fields: - `info`: metadata (description, version, year) - `license`: licensing information - `images`: list of image entries with `id`, `file_name`, `height`, `width` - `annotations`: list of annotations, each containing: - `id`: unique annotation ID - `image_id`: references the corresponding image - `category_id`: always 1 (golden_monkey) - `keypoints`: a flat array of length 51 (17×3) - `num_keypoints`: number of annotated keypoints - `bbox`: [x, y, width, height] in pixel coordinates - `area`: area of the bounding box - `segmentation`: empty (polygon not used) - `categories`: category definition (including keypoint names and skeleton)