Can this work generalize to half-portrait image?

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

710 stars 130 forks source link

Can this work generalize to half-portrait image? #40

Closed hot-dog closed 3 years ago

hot-dog commented 3 years ago

I tested snapshot_8.pth.tar with a half-portrait image which only contains torso and the above parts, it seems that the output_mesh_lixel.obj is also half-portrait and i think it should be correct, but the output_mesh_param.obj is full-portrait, it produces extra hip and legs part. Any suggestion? Thank you!

mks0601 commented 3 years ago

As lixel-based heatmap is defined in the image space, it cannot recover area beyond images. On the other hand, param-based meshes are defined in 3D space. Therefore, it can always recover full body shape.

hot-dog commented 3 years ago

Thanks for your reply, i get it.