| --- |
| license: cc-by-nc-4.0 |
| language: |
| - en |
| pretty_name: PanoMouse |
| tags: |
| - mouse |
| - pose-estimation |
| - behavior-analysis |
| - neuroscience |
| - multiview system |
| task_categories: |
| - keypoint-detection |
| - video-classification |
|
|
| configs: |
| - config_name: default |
| data_files: |
| - split: train |
| path: "train_annotations_2d.jsonl" |
| - split: test |
| path: "test_annotations_2d.jsonl" |
| --- |
| # PanoMouse Dataset |
|
|
| PanoMouse is the official dataset for the paper: **Mouse Total Capture: A Multi-view Dataset for 3D Motion and Expression Capture of Freely Moving Mouse** |
|
|
| PanoMouse supports simultaneous whole-body motion and facial expression capture |
| of freely moving mice via a 24-camera multiview system. |
|
|
| ## Dataset Overview |
|
|
| - **Cameras**: 24 cameras distributed across top-middle-bottom 3 layers, 360° coverage |
| - **Frames**: Over 3.4 million frames |
| - **Annotations**: 7,293 frames annotated with **92 whole-body keypoints** |
| - **Tasks**: 2D pose estimation, 3D pose estimation, behavior classification |
|
|
| ## Repository Structure |
| ``` |
| PanoMouse/ |
| ├── action_annotations/ # Action category labels for behavior classification |
| ├── annotations/ # 2D and 3D pose annotations (keypoint coordinates) |
| ├── videos-1K.zip # Multi-view synchronized videos (resize to 1K resolution) |
| └── images.zip # Multi-view image frames to train 2D pose and 3D pose models |
| ``` |
|
|
