Hi
Thank you for releasing the code for us. Now I'm using your network on my own dataset. It's a eye-track dataset and each 3d triangle mesh have vertices and corresponding RGB value(it can be regarded as saliency value of this vertex). And I have already downsampled the 3d mesh to 1024 point cloud and set the label of each vertex = 2, if the vertex is fixated, otherwise 1(I'm not sure whether it's 1-based integer label or not). Another question is the obj file after test. I see this file have many vertices but I can't visualize it using meshlab. Could you please give me some explanations or suggestions about how can I generate a saliency map like what you do in pointgrid paper figure4?
Hi Thank you for releasing the code for us. Now I'm using your network on my own dataset. It's a eye-track dataset and each 3d triangle mesh have vertices and corresponding RGB value(it can be regarded as saliency value of this vertex). And I have already downsampled the 3d mesh to 1024 point cloud and set the label of each vertex = 2, if the vertex is fixated, otherwise 1(I'm not sure whether it's 1-based integer label or not). Another question is the obj file after test. I see this file have many vertices but I can't visualize it using meshlab. Could you please give me some explanations or suggestions about how can I generate a saliency map like what you do in pointgrid paper figure4?