3D-BoNet icon indicating copy to clipboard operation
3D-BoNet copied to clipboard

Visualization of res_by_scene to point cloud and masks

Open ninasinger opened this issue 6 years ago • 5 comments

I am interested in visualizing the point masks for an entire scene, but I am a little confused by the format of the .mat file. Do you have any code that would out put the .mat file in res_by_scene to a .txt file or other point cloud type file? I am able to visualize point cloud masks for each batch but am having trouble doing the same for the entire scene. I am looking for a visualization similar to Fig. 7 in your paper. Thanks in advance

ninasinger avatar Mar 03 '20 17:03 ninasinger

I am interested in visualizing the point masks for an entire scene, but I am a little confused by the format of the .mat file. Do you have any code that would out put the .mat file in res_by_scene to a .txt file or other point cloud type file? I am able to visualize point cloud masks for each batch but am having trouble doing the same for the entire scene. I am looking for a visualization similar to Fig. 7 in your paper. Thanks in advance

To author: I also meet the same problem. Would you mind offering the solving method ? Thanks in advance !

ZJU-PLP avatar Mar 05 '20 01:03 ZJU-PLP

Hi @maptastic @ZJU-PLP , it's very easy to save the results of each scene, just uncomment the following lines to check the results are correct and then save the variables like "ins_pred_all", "sem_pred_all", etc. to whatever format you prefer.

https://github.com/Yang7879/3D-BoNet/blob/37bf0be9de618ecd69e0138ee75d8d573a0fcea1/main_eval.py#L212

Yang7879 avatar Mar 06 '20 15:03 Yang7879

success! thank you

ninasinger avatar Mar 13 '20 20:03 ninasinger

Can you also point me towards the same info for the bounding boxes? I am looking for the bounding boxes per scene instead of by batch

ninasinger avatar Mar 17 '20 20:03 ninasinger

Hi @maptastic , there might be some algorithms to merge the predicted bounding boxes according to the overlap, but I guess the bounding boxes per scene can be obtained once you have the object instance results per scene (can just use the min-max vertices to roughly draw a bounding box).

Yang7879 avatar Mar 18 '20 12:03 Yang7879