search

magic-research / GETAvatar

[ICCV 2023] GETAvatar: Generative Textured Meshes for Animatable Human Avatars
https://getavatar.github.io/
MIT License
106 stars 8 forks source link

Train on other datasets. #12

Open somnus1840 opened 11 months ago

somnus1840 commented 11 months ago

If I want to train the model on other non 3D scanning format datasets such as deepfashion, do I need to obtain SMPL labels in the following format: label info: 0:16 world2camera_matrix shape (16,) 16:19 global_orient shape (3,) 19:88 body_pose shape (69,) 88:98 betas shape (10,) """ If so, could you introduce me a work that can obtain SMPL parameters in this format? Thank you very much!

Xuanmeng-Zhang commented 7 months ago

Hi, you can refer to PARE to obtain approximate camera and SMPL parameters, and the prepare them into the format.