xubin1994

> IIRC "extremely hard" samples have disparity magnitudes > 300 pixels (I am not sure whether there was also a flow criterion, but I do not think so). I think...

> Hi, > > Some samples in the dataset are used for debugging, and cannot be used for network training. You may find the lists of all the valid training...

> I see. It is OK to remove those samples when you train/test your model. > Since the data set is large, we need time to check all of them....

Hi,can you provide the correct train list file now? There are still some wrong examples in the training list,such as ''IRSDataset/Home/ModernHotelRoom/l_31.png'',the left image and the disparity image are shifted.

> Hi, I have already installed mmdeploy from **git clone [[email protected]](mailto:[email protected]):drilistbox/mmdeploy.git**, but there is an error when I used the command: python tools/convert_bevdet_to_TRT.py $config $checkpoint $work_dir --fuse-conv-bn --fp16 Could you...