Open ZJU-PLP opened 1 year ago
@aragakiyui611 Hi, dear author: When I want to reproduce the visualization results of BiCo-Net, I meet one problem. Could you mind helping me to solve this issue?(I also have sent an email to your mailbox).
I cannot get the right visualization results. Specially, the prediction point cloud cannot be visualized right while the target point cloud can do.
target_t = target_t + centroid out_t = out_t + centroid target = torch.bmm(model_points, target_r.transpose(2, 1)) + target_t # (1, 2620, 3) pred = torch.bmm(model_points, out_R.transpose(2, 1)) + out_t # (1, 2620, 3)
Visualization results: example 1: example 2:
Hello, I also encountered the problem of not being able to visualize, could you please share your code? Hope to study together.
@aragakiyui611 Hi, dear author: When I want to reproduce the visualization results of BiCo-Net, I meet one problem. Could you mind helping me to solve this issue?(I also have sent an email to your mailbox). I cannot get the right visualization results. Specially, the prediction point cloud cannot be visualized right while the target point cloud can do.
target_t = target_t + centroid out_t = out_t + centroid target = torch.bmm(model_points, target_r.transpose(2, 1)) + target_t # (1, 2620, 3) pred = torch.bmm(model_points, out_R.transpose(2, 1)) + out_t # (1, 2620, 3)
Visualization results: example 1: example 2:
Hello, I also encountered the problem of not being able to visualize, could you please share your code? Hope to study together.
I am very sorry that I have not solved this problem. Moreover, you can try to use the script code eval_ycb.py in object-posenet to visualize the results after saving object predicting poses.
@aragakiyui611 Hi, dear author: When I want to reproduce the visualization results of BiCo-Net, I meet one problem. Could you mind helping me to solve this issue?(I also have sent an email to your mailbox). I cannot get the right visualization results. Specially, the prediction point cloud cannot be visualized right while the target point cloud can do.
target_t = target_t + centroid out_t = out_t + centroid target = torch.bmm(model_points, target_r.transpose(2, 1)) + target_t # (1, 2620, 3) pred = torch.bmm(model_points, out_R.transpose(2, 1)) + out_t # (1, 2620, 3)
Visualization results: example 1: example 2:
Hello, I also encountered the problem of not being able to visualize, could you please share your code? Hope to study together.
I am very sorry that I have not solved this problem. Moreover, you can try to use the script code eval_ycb.py in object-posenet to visualize the results after saving object predicting poses.
ok, thanks for your reply.
@aragakiyui611 Hi, dear author: When I want to reproduce the visualization results of BiCo-Net, I meet one problem. Could you mind helping me to solve this issue?(I also have sent an email to your mailbox).
I cannot get the right visualization results. Specially, the prediction point cloud cannot be visualized right while the target point cloud can do.
Visualization results: example 1: example 2: