DeepLabCut Model
Published by Hilda Azimi on ctlut (s3://bucket/path or a data asset ID).
Keypoints
36 keypoints: eye_top_l, eye_bottom_l, pupil_left_l, pupil_right_l, pupil_top_l, pupil_bottom_l, whisker_pad_l, tube_bottom, tongue_end_l, tube_top, tongue_end_r, chin_l, pinky_l, ring_finger_l, middle_finger_l, pointer_finger_l, pinky_lh, ring_finger_lh, middle_finger_lh, pointer_finger_lh, ear_tip_l, whisker_pad_r, chin_r, pinky_r, ring_finger_r, middle_finger_r, pointer_finger_r, pinky_rh, ring_finger_rh, middle_finger_rh, pointer_finger_rh, ear_tip_r, nostril_l, nostril_r, nose_tip, tail_tip
Performance
Pixel error = mean Euclidean distance (px) between predicted and ground-truth keypoints. Lower is better.
| Set | Error (px) | Error w/ p-cutoff (px) |
|---|---|---|
| Train | 0.93 | 0.93 |
| Test | 6.18 | 2.51 |
training iterations: 1030000; p-cutoff: 0.6.
Changelog
v1.0
what this model is / what changed
- Downloads last month
- 15