search

leoxiaobin / deep-high-resolution-net.pytorch

The project is an official implementation of our CVPR2019 paper "Deep High-Resolution Representation Learning for Human Pose Estimation"
https://jingdongwang2017.github.io/Projects/HRNet/PoseEstimation.html
MIT License
4.32k stars 913 forks source link

Example code or tutorial on using models #158

Open jwayne2978 opened 4 years ago

jwayne2978 commented 4 years ago

Is there code on just using the models for inference? I want to feed the model images (each image may have multiple humans), and I would expect the outputs to be the body part estimations (in JSON or whatever format).

As I am reading the issues, especially the one on making inference with a web camera, I "think" that using these models is not trivial. For example, it seems to be I have to detect "people" (use an object detection model, for example), and then supply the center of the bounding box to the model to predict the pose components. Is this correct? This approach would seem like a multi-step process.

alex9311 commented 4 years ago

I would say this issue can be closed with #161 being merged