search

Pose-Group / DCPose

This is an official implementation of our CVPR 2021 paper "Deep Dual Consecutive Network for Human Pose Estimation" (https://openaccess.thecvf.com/content/CVPR2021/papers/Liu_Deep_Dual_Consecutive_Network_for_Human_Pose_Estimation_CVPR_2021_paper.pdf)
380 stars 61 forks source link

hi ,bro.how can i run code using my own video? #8

Closed qhdqhd closed 3 years ago

qhdqhd commented 3 years ago 
cd demo/                   
mkdir input/
# Put your video in the input directory
python video.py

Your method above didnt work.Maybe because the pretrained model (pretrained_coco_model)is not suitable? thanks a lot !

chenhaomingbob commented 3 years ago

Hi qhdqhd : I've tested my code and it works fine. Posting your error message and I will help you.

qhdqhd commented 3 years ago

Thank you,chenhaomingbob: (DCPose) qhd@qhd-Lenovo:~/0code/pose/DCPose-main/demo$ python video.py Traceback (most recent call last): File "video.py", line 13, in <module> from tools.inference import inference_PE File "/home/qhd/0code/pose/DCPose-main/tools/inference.py", line 64, in <module> model = get_inference_model() File "/home/qhd/0code/pose/DCPose-main/tools/inference.py", line 58, in get_inference_model model_state_dict = {k.replace('module.', ''): v for k, v in checkpoint_dict['state_dict'].items()} KeyError: 'state_dict' I use pretrained_coco_model.pth in google drive https://drive.google.com/drive/folders/1RYwOsNDsCZAsJe44MnpwHmwA3GSTDcAr

qhdqhd commented 3 years ago

@chenhaomingbob this is my error, it seems that the pretrained model is not suitable ?

chenhaomingbob commented 3 years ago

Hi @qhdqhd: You should use pre-trained file: PoseTrack17_DCPose.pth. As the following code is written: https://github.com/Pose-Group/DCPose/blob/078f4495c654f7220fa599fda07a8eef5dc54f21/tools/inference.py#L33

qhdqhd commented 3 years ago

Thankyou, I find PoseTrack17_DCPose.pth just now, it is a little unremarkable. @chenhaomingbob