{"success":true,"data":{"author":"Sawtell Lab and Paninski Lab, Columbia University","bucketAccessType":"public","bucketName":"paninski-lab/mirror-fish","bucketPrefix":"","category":"Neuroscience","cloudProvider":"huggingface","createdAt":"2026-11-21T20:45:58.741Z","description":"A labeled animal pose estimation dataset of weakly electric mormyrid fish (elephantnose fish, Gnathonemus petersii) swimming freely and capturing worms, recorded in the Sawtell Lab at Columbia University by Federico Pedraja, David Ehrlich and Dillon Noone. A single camera at 300 Hz sees a direct view plus two 45 degree mirror views (top and side), so each frame is labeled with 51 keypoints: 17 body parts (chin, head, pectoral fins, dorsal and anal fins, caudal fin, fork) in three views. 373 in-distribution frames from 28 sessions and 94 out-of-distribution frames from 10 different sessions and animals. Packaged by the Paninski Lab for Lightning Pose: labeled-data/ PNG frames with context frames, CollectedData.csv and CollectedData_test.csv (DeepLabCut-style 2D labels), videos, 51-frame clips around each test frame for the Ensemble Kalman Smoother, a training config and project.yaml. Source archive: Lightning Pose dataset mirror-fish, DOI 10.6084/m9.figshare.24993363, CC BY 4.0. Cite that record and the Lightning Pose paper (Biderman et al., Nature Methods 2024).","displayTitle":"Mirror Fish: Lightning Pose Keypoint Dataset of Weakly Electric Mormyrid Fish (Gnathonemus petersii) Filmed with Mirrors at 300 Hz (Columbia University)","doi":"10.6084/m9.figshare.24993363","huggingFaceUrl":"https://huggingface.co/datasets/paninski-lab/mirror-fish","id":"8e6d4775-2dec-5c19-a3f2-4ce70b9ed9c9","license":"CC-BY-4.0","licenseEvidenceUrl":"https://api.figshare.com/v2/articles/24993363","objectCountEstimate":2473,"probeStatus":"ok","referenceUrl":"https://huggingface.co/datasets/paninski-lab/mirror-fish","region":"us","schemaCard":{"accessHints":[{"language":"python-datasets","snippet":"from datasets import load_dataset\nds = load_dataset(\"paninski-lab/mirror-fish\")"},{"language":"python-hub","snippet":"from huggingface_hub import snapshot_download\nsnapshot_download(repo_id=\"paninski-lab/mirror-fish\", repo_type=\"dataset\")"}],"caveats":["HuggingFace metadata: author=paninski-lab, license:cc-by-4.0, size_categories:n<1K, downloads=91"],"format":"mixed","generatedAt":"2026-11-21T20:48:46.769Z","layout":{"example":"CollectedData.csv","pattern":"{repoId}/{filename}"},"mountPath":"/mnt/archive","sampleHeaders":[{"bytes":"scorer,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,rick,","encoding":"utf8","key":"CollectedData.csv","truncatedAt":512}],"sampleObjects":["CollectedData.csv","CollectedData_test.csv","config_mirror-fish.yaml","labeled-data/20200909_Fred/img000685.png","labeled-data/20200909_Fred/img000686.png"],"sidecars":[{"bytes":"---\ntask_categories:\n  - keypoint-detection\nlicense: cc-by-4.0\ntags:\n  - biology\n  - pose-estimation\n  - fish\n  - lightning-pose\npretty_name: Mirror Fish\nsize_categories:\n  - n<1K\n---\n\n# Mirror Fish — Lightning Pose Single-View Dataset\n\nSingle-camera pose estimation dataset for mormyrid fish body keypoints, packaged for use with [Lightning Pose](https://github.com/paninski-lab/lightning-pose).\n\n## Dataset Description\n\nWeakly-electric mormyrid fish (*Gnathonemus petersii*) swim freely in and out of an experimental tank, capturing worms from a well. The tank has a side mirror and a top mirror, both at 45°, allowing a single camera to capture three views simultaneously — a direct view and two mirror views — at **300 Hz**. Each frame is labeled with **51 keypoints**: 17 body parts across all three views.\n\nSource data: original archive at https://doi.org/10.6084/m9.figshare.24993363.\nData collected by Federico Pedraja, David Ehrlich, and Dillon Noone in the Sawtell Lab, Columbia University.\n\n## Data Splits\n\n| Split | Labeled frames | Sessions |\n|-------|---------------:|---------:|\n| In-distribution (InD) | 373 | 28 |\n| Out-of-distribution (OOD) | 94 | 10 |\n\nInD and OOD sets contain **different sessions / animals** (no overlap).\n\n- `CollectedData.csv` — InD labels; `videos/` — InD videos\n- `CollectedData_test.csv` — OOD labels; `videos_test/` — OOD videos\n\n## Keypoints\n\n51 keypoints total: 17 body parts × 3 views (`_main`, `_top`, `_right`).\n\n| Body part | Main view | Top view | Right view |\n|-----------|-----------|----------|------------|\n| Chin tip | chin_tip_main | chin_tip_top | chin_tip_right |\n| Chin ¾ | chin3_4_main | chin3_4_top | chin3_4_right |\n| Chin half | chin_half_main | chin_half_top | chin_half_right |\n| Chin ¼ | chin1_4_main | chin1_4_top | chin1_4_right |\n| Chin base | chin_base_main | chin_base_top | chin_base_right |\n| Head | head_main | head_top | head_right |\n| Mid | mid_main | mid_top | mid_right |\n| Tail neck | tail_neck_main | tail_neck_top | tail_neck_right |\n| Caudal ventral | caudal_v_main | caudal_v_top | caudal_v_right |\n| Caudal dorsal | caudal_d_main | caudal_d_top | caudal_d_right |\n| Pectoral L base | pectoral_L_base_main | pectoral_L_base_top | pectoral_L_base_right |\n| Pectoral L | pectoral_L_main | pectoral_L_top | pectoral_L_right |\n| Pectoral R base | pectoral_R_base_main | pectoral_R_base_top | pectoral_R_base_right |\n| Pectoral R | pectoral_R_main | pectoral_R_top | pectoral_R_right |\n| Dorsal | dorsal_main | dorsal_top | dorsal_right |\n| Anal | anal_main | anal_top | anal_right |\n| Fork | fork_main | fork_top | fork_right |\n\n## Directory Structure\n\n```\nmirror-fish/\n├── labeled-data/                    # Extracted frames per session; includes ±2 context frames\n├── videos/                          # InD session video clips\n├── videos_test/                     # OOD session video clips\n├── videos-for-each-labeled-frame/   # 51-frame videos centered on each OOD labeled frame\n├── CollectedData.csv                # InD 2D keypoint labels (x,y per keypoint)\n├── CollectedData_test.csv           # OOD 2D keypoint labels\n├── config_mirror-fish.yaml          # Sample Lightning Pose training config\n└── project.yaml                     # View and keypoint definitions (required by LP App)\n```\n\nThe `videos-for-each-labeled-frame/` directory contains 51-frame video clips with the labeled frame at the center, intended for use with temporal smoothers such as the [Ensemble Kalman Smoother](https://github.com/paninski-lab/eks).\n\nSee the Lightning Pose documentation for full details on the [single-view data directory structure](https://lightning-pose.readthedocs.io/en/latest/source/directory_structure_reference/singleview_structure.html).\n\n## Usage with Lightning Pose\n\nThe included `config_mirror-fish.yaml` is a ready-to-use training config. Key settings:\n\n- **Image resize:** 256 × 384\n- **Backbone:** `resnet50_animal_ap10k`\n- **Keypoints:** 51\n- **Mirror columns:** `[0–16]` (main), `[17–33]` (top), `[34–50]` (right)\n\nUpdate `data.","encoding":"utf8","key":"README.md","name":"README.md","size":4999,"truncated":true}],"sourceObjectKey":"CollectedData.csv","starterCell":"import pandas as pd\ndf = pd.read_csv(\"hf://datasets/paninski-lab/mirror-fish/CollectedData.csv\")\ndf.head()","starterSamplePath":"CollectedData.csv","totalObjects":2473},"summary":"A labeled animal pose estimation dataset of weakly electric mormyrid fish (elephantnose fish, Gnathonemus petersii) swimming freely and capturing worms, recorded in the Sawtell…","title":"Mirror Fish: Lightning Pose Keypoint Dataset of Weakly Electric Mormyrid Fish (Gnathonemus petersii) Filmed with Mirrors at 300 Hz (Columbia University)","updatedAt":"2026-11-22T00:57:11.606Z","visibility":"public","websiteUrl":"https://github.com/paninski-lab/lightning-pose"}}