PointRCNN Testing on different data

Hi, I have a different data set with just the point cloud information. I wanted to know how would I go about to test your model on the data that I have. What are the changes that I need to make for this model to work on that data. What is the command that I need to use? I do not have any camera data.

Sep 26 '19 15:09 PrithviKambhampati

You can follow the code and update in the eval_rcnn part and use the cam axis definition Velodyne points to feed your model if you're going to use a KITTI pretrained model for inference. Can use the rounded version of KITTI calibration for V2C if your Velodyne data has the same axis definitions as KITTI Velodyne.

Oct 01 '19 10:10 bmankirLinker

@PrithviKambhampati My suggestions would be rewrite a dataloader, if you're still confuesed, maybe you could reference my closed PR and in /dataset/lytf_dataset.py, where i def a new dataloader for only lidar point cloud and labels input

Oct 11 '19 16:10 DaHaiHuha

@PrithviKambhampati My suggestions would be rewrite a dataloader, if you're still confuesed, maybe you could reference my closed PR and in /dataset/lytf_dataset.py, where i def a new dataloader for only lidar point cloud and labels input

thank you for your suggestioon, but can you share your a new dataloader code for only lidar point cloud and labels input

Jan 13 '20 01:01 yinhai86924