Attention3DHumanPose
Attention3DHumanPose copied to clipboard
How to evaluate custom video?
Could you please tell me how to run your code with custom video
Hello, have you solved the problem? I also want to evaluate custom video.
I’m sorry to say that I didn’t solve the problem
If the pipeline is consistent with VideoPose3D, you can refer to Inference in the wild to infer in the wild. But you will need matched 2D model and 3D model, like 2D model outputs coco format 2D keypoints and 3D takes coco format 2D keypoints as input or 2D model outputs h36m format 2D keypoints and 3D takes h36m format 2D keypoints as input.