Visualization of fit_joint_cam

mks0601 / I2L-MeshNet_RELEASE

Official PyTorch implementation of "I2L-MeshNet: Image-to-Lixel Prediction Network for Accurate 3D Human Pose and Mesh Estimation from a Single RGB Image", ECCV 2020

MIT License

708 stars 130 forks source link

Visualization of fit_joint_cam #93

Closed Bozcomlekci closed 2 years ago

Bozcomlekci commented 2 years ago

Greetings, I'm trying to visualize the joints via fit_joint_cam. I use vis_keypoints(img, kps, alpha=1) function in common/utils/vis.py. I get empty image results. Am I doing something wrong? Could you tell the input shapes for the arguments of the function (img and kps). Can I input fit_joint_cam as kps?

mks0601 commented 2 years ago

fit_joint_cam are defined in camera-centered coordinate system, which has a milimeter (or meter) unit. You should project them to image space before trying to visualize them. The projected one is fit_joint_img.

Bozcomlekci commented 2 years ago

Thanks for the explanation, I have corrected it.