search

HongwenZhang / PyMAF-X

[TPAMI 2023] PyMAF-X: Towards Well-aligned Full-body Model Regression from Monocular Images
https://www.liuyebin.com/pymaf-x
Other
196 stars 27 forks source link

Training process (and congrats TPAMI accepting) #15

Open damonchang23 opened 1 year ago

damonchang23 commented 1 year ago

First of all, congrats accept TPAMI your great work. I want to training your network from sketch for research purpose but there is no training precedure or details in your README.md. If you don't mind, can you guide how to train your network from sketch?

Thank you for your time and help.

HongwenZhang commented 1 year ago

Hi, sorry for the slow update of this project.

To train the code, the most tedious part is the preparation of the training data. I have just uploaded the compatible SMPL-X labels, and hope these files could help.

The example script to run the training code: python -m apps.train --regressor pymaf_net --train_data h36m_coco_itw --pretrained_checkpoint xxxx --misc TRAIN.BATCH_SIZE 64 TRAIN.NUM_WORKERS 8 LOSS.SHAPE_W 0.6 MODEL.PyMAF.BACKBONE hr48

fatbao55 commented 1 year ago

Hi @HongwenZhang, thanks for the great work! Do you also mind sharing your training process and data preparation for the AGORA benchmark?

ruowenyu commented 11 months ago

Hi, @damonchang23, I also wand to train PyMAF-X from scatch, but encounter some issues during training. I was wondering if you train PyMAF-X successfully? Thanks for the help!