Bo Yang

Hi @Violetit , I understand that TensorFlow is a bit painful for beginners, but after one or two months, you will feel it's pretty easy and flexible to use. Here...

Hi @LeeLinSCUT, the hyperparameters are the same as what I used to train, but the results may indeed be different (can be higher or lower), probably because the association algorithm...

hi @longmalongma, to visualize either semantic or instance labels, check out this function: https://github.com/Yang7879/3D-BoNet/blob/07c99e58f7f1a646c211c150cb7b3c363fdd69d9/helper_data_plot.py#L37

hi @longmalongma, since s3dis dataset does not have faces for individual points, it's not straightforward to save the results as meshes. For your benefit, here's a script to write a...

Hi @pomelo93 , here it is. 链接:https://pan.baidu.com/s/1ww_Fs2D9h7_bA2HfNIa2ig 密码:qpt7

hi @masotrix, you may double-check whether all blocks (i.e., each individual point cloud u feed into the network) contain at least a valid instance. If there is no valid instance,...

Hi @masotrix , for each point cloud (say 4096 points), there is at least one instance inside, note that an instance may not necessarily be a complete object, it could...

hi @zhangguanghui1, you need to train/test 6 models in total, i.e., (1) train on Area23456, test on Area1, ..., (6) train on Area12345, test on Area6. The final results are...

Hi @Violetit, thanks for your interest in our work. Basically the code of the netwok (including all configurations) for scannet is the same as the released code used for s3dis....

hi @cuge1995 , you are suggested to compile the tf_ops correctly before training the network.