Bo Yang

Hi @cffan, thank you for your interest in our work. We are still improving it, which might take a couple of weeks. If you are in urgent need, you are...

Hi @lin1061991611 , exactly as you said.

Hi @abhigoku10 , for the time being, we haven't trained the model on other datasets yet, but I believe it is able to be trained if you follow a similar...

@abhigoku10 , the simplest approach is to randomly duplicate some points for upsampling.

Hi @lifeiwen, it seems that some value(s) of the cost matrix is nan? this numerical issue may happen when computing the costs (siou or ce).

hi @neshatbln , it seems the dataset path is not correct??? hope you've solved it already.

hi @lin1061991611 , "volume" merges the instance labels, while "volume_seg" tracks the semantic labels.

Hi @fxy2012 , tbh, I don't know how to make it work on a windows machine. If you need to visualize the pc, I suggest you use Matlab in windows...

@fxy2012 , (1) Plz check whether the sem gt labels are well loaded from your datasets. (2) After obtaining the instance segmentation results, you may compute the bounding box by...

@fxy2012, For each category, the Prec is computed after gathering predicted results from all scenes. The mPrec is the average score from all categories. Plz check the code below carefully....