The performance in Shelf

[baojunshan]

I used Shelf trained model to evaluate the Shelf, and get the pose results in

PlaneSweepPose/lib/dataset/shelf.py

Line 332 in 35d8bc3

pred = np.stack([self.coco2shelf3D(p[:, :3]) for p in final_pose3d_pool]) # [Np, Nj, 3]

, and then visualized it showed above.

The result shows a lot of noise, I don't know if this results is the same as paper shows, or I implemented incorrectly.

[jiahaoLjh]

@baojunshan

If a person is not visible in multiple views, the depth cannot be estimated correctly because of wrong multi-view association. In that case, we will observe those noisy poses far from the scene center of the multi-view camera system.

[UsamaShaikh1]

Hello @baojunshan, do you still have the visualization code for the above 3D film and if so can you please share it with me