search

facebookresearch / sapiens

High-resolution models for human tasks.
https://about.meta.com/realitylabs/codecavatars/sapiens/
Other
3.91k stars 199 forks source link

get human body keypoints only when use the '308' model #53

Closed luoshuiyue closed 2 weeks ago

luoshuiyue commented 2 weeks ago

hello~ there are points stored in a json file (run pose estimation), May I ask how to acquire only the 17 human body key points when use the '308 keypoints' model?

luoshuiyue commented 2 weeks ago

and how do I use lines to connect key points, as in the example you provided (the second picture): 1746ce7c77a5d27e994da214a4559e6 67f9f092da77eaa82b7e3d76a94da31

luoshuiyue commented 2 weeks ago

when I follow full installation, I encounter this error: 1725356424557

rawalkhirodkar commented 2 weeks ago

@luoshuiyue To get the 17 body keypoints following COCO, index into the predicted 308 keypoints using this

The indices for the 17 COCO keypoints are: 0: dict(name='nose', id=0, color=[51, 153, 255], type='upper', swap=''), 1: dict(name='left_eye', id=1, color=[51, 153, 255], type='upper', swap='right_eye'), 2: dict(name='right_eye', id=2, color=[51, 153, 255], type='upper', swap='left_eye'), 3: dict(name='left_ear', id=3, color=[51, 153, 255], type='upper', swap='right_ear'), 4: dict(name='right_ear', id=4, color=[51, 153, 255], type='upper', swap='left_ear'), 5: dict(name='left_shoulder', id=5, color=[51, 153, 255], type='upper', swap='right_shoulder'), 6: dict(name='right_shoulder', id=6, color=[51, 153, 255], type='upper', swap='left_shoulder'), 7: dict(name='left_elbow', id=7, color=[51, 153, 255], type='upper', swap='right_elbow'), 8: dict(name='right_elbow', id=8, color=[51, 153, 255], type='upper', swap='left_elbow'), 9: dict(name='left_hip', id=9, color=[51, 153, 255], type='lower', swap='right_hip'), 10: dict(name='right_hip', id=10, color=[51, 153, 255], type='lower', swap='left_hip'), 11: dict(name='left_knee', id=11, color=[51, 153, 255], type='lower', swap='right_knee'), 12: dict(name='right_knee', id=12, color=[51, 153, 255], type='lower', swap='left_knee'), 13: dict(name='left_ankle', id=13, color=[51, 153, 255], type='lower', swap='right_ankle'), 14: dict(name='right_ankle', id=14, color=[51, 153, 255], type='lower', swap='left_ankle'), 41: dict(name='right_wrist', id=41, color=[51, 153, 255], type='upper', swap='left_wrist'), 62: dict(name='left_wrist', id=62, color=[51, 153, 255], type='upper', swap='right_wrist'),

rawalkhirodkar commented 2 weeks ago

@luoshuiyue for the numpy error with xtcocotools, you will need to modify the xtcocotools slightly to be compatible with new numpy or alternatively downgrade numpy version.

More details here.

luoshuiyue commented 2 weeks ago

@luoshuiyue for the numpy error with xtcocotools, you will need to modify the xtcocotools slightly to be compatible with new numpy or alternatively downgrade numpy version.

More details here.

@rawalkhirodkar hi~I downgrade numpy version to 1.26.4, this error was gone but a new one appears: 微信图片_20240905113944

rawalkhirodkar commented 2 weeks ago

@luoshuiyue Thanks for sharing the log, in hindsight, downgrading numpy is not a good idea.

Following the link I shared, can you try pip install git+https://github.com/jin-s13/xtcocoapi? Let me know who it goes.