Naman-ntc / Pytorch-Human-Pose-Estimation

Implementation of various human pose estimation models in pytorch on multiple datasets (MPII & COCO) along with pretrained models
MIT License
474 stars 75 forks source link

where is train.h5? #12

Closed electronicliujiang closed 5 years ago

electronicliujiang commented 5 years ago

thanks for your code!! I can't find your train.h5. Where I can find this? in your model.zip?

electronicliujiang commented 5 years ago

I have found this thank you!!

electronicliujiang commented 5 years ago

How to achieve visualization?For example, key point detection visualization of pictures and videos?

Naman-ntc commented 5 years ago

Hi, thanks for interest in this work and sorry I do not have the visualization code formatted and arranged properly so I can't gaurentee its correctness. However you can find this code and suit it to your requirements!

electronicliujiang commented 5 years ago

Hello my friend! I have some question. Currently, I have been able to implement some visual operations, but they are all based on the mpii dataset. I would like to ask, how to check the key points on my own data set, my data set does not have s (scale) and c (center), how to set these two parameters?

------------------ 原始邮件 ------------------ 发件人: "Naman Jain"notifications@github.com; 发送时间: 2019年10月4日(星期五) 晚上8:12 收件人: "Naman-ntc/Pytorch-Human-Pose-Estimation"Pytorch-Human-Pose-Estimation@noreply.github.com; 抄送: "657361488"657361488@qq.com;"Author"author@noreply.github.com; 主题: Re: [Naman-ntc/Pytorch-Human-Pose-Estimation] where is train.h5?(#12)

Hi, thanks for interest in this work and sorry I do not have the visualization code formatted and arranged properly so I can't gaurentee its correctness. However you can find this code and suit it to your requirements!

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.

Naman-ntc commented 5 years ago

Hi, I don't think there is any easy option. In case you have gt. skeleon coordinates you can develop rought center and scale of person but otherwise you will need a trained detector to get rough annotations for your dataset