Pose Tracking
Collection
This collections hosts a set of pose tracking models developed at the Allen Institute • 2 items • Updated
Lightning Pose Model
Trained by Hilda Azimi on ctlut.
- Backbone:
resnet50_animal_ap10k - Losses used:
['pca_singleview'] - Keypoints (9):
tube_left, tube_right, nose_tip, jaw_tip, jaw_left, jaw_right, tongue_tip, tongue_left, tongue_right - Epochs: 300-300, train/val split: 0.95/0.05
- Supervised RMSE (validation): 1.78 resized-px
Performance
Supervised RMSE (from TensorBoard logs, unit: resized-px — measured on the resized training image, not original-frame pixels). Lower is better.
No per-keypoint pixel-error file was found; the numbers below come from the training logs instead.
Source: events.out.tfevents.1770319508.8a5308b38484.94.0
| Set | Final RMSE | Best RMSE |
|---|---|---|
| train | 0.36 | 0.26 |
| validation | 1.95 | 1.78 |
Changelog
v1.0
what this model is / what changed
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support