Mimi Liao

Hi, thank you for your response. I used mini-batch size 64 and num_threads 16, which is totally the same as you. I only used one RTX3090 but it doesn't have...

So if I used only one gpu, should I divide my learning rate by 4?

Also, Does the global rotation mentioned in your paper refer to the root rotation with respect to the world coordinate or to the camera coordinate?

I see your supp, and understand the above questions. However, in the below picture, if now we used the cropped image as camera coordinate(the original version), should the tx and...

I changed my config file as below: ``` trainset_3d: ['Human36M', 'MuCo'] trainset_2d: ['MSCOCO', 'MPII'] testset: 'PW3D' lr_dec_epoch: [30] end_epoch: 40 lr: 0.0005 lr_backbone: 0.0005 lr_dec_factor: 10 # modify batch size...

1. I evaluate on 3DPW. not Crowd. 2. I used RTX3090 with batch size 128. The training time is 0.91h / epoch.

So sorry that I pasted the wrong command. My training command is `python train.py --amp --gpu 0 --cfg ../assets/yaml/3dpw.yml` This is my full code: https://github.com/mimiliaogo/3DCrowdNet-Mimi Thank you so much!

Hi, I tried your conifg as 3., the results seem normal. So maybe the problem is from training data. I will try to visualize them. BTW, there is no change...

@hongsukchoi, when I train your model with Human3.6M and MuCo respectively, both of them will have increasing errors. I visualize the GT keypoints and joints, and the results seem normal...

@hongsukchoi Thank you for getting back to me. But I used your model weight given in [here](https://drive.google.com/drive/folders/19ntGuC0zaXQa3cCN_2Ox_hWYX3nLLP2J ) with version 04-06 snapshot 10, and evaluated it on Human3.6M but got...