Baturalp Mankır

@shengyuwoo since you don't provide the commands, I can't understand what you were really up to :)

My guess is that it looks like your training in the second stage had some issues, RPN side didn't load correctly so the final ` ckpt ` RCNN has missing...

> > My guess is that it looks like your training in the second stage had some issues, RPN side didn't load correctly so the final `ckpt` RCNN has missing...

> > @shengyuwoo ,I have the same problem as yours. And I tried to run the second line of inference code.I found that in ../output/rcnn/ckpt folder , checkpoint_epoch_32.pth can test...

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...

You can try to evaluate the `rpn` models before selecting the `checkpoint_epoch_200.pth`.

there are some tips in the PointPillars paper, https://arxiv.org/pdf/1812.05784.pdf .

Those are the ignored values which are set in `DataLoader` for nearby points of the enlarged cuboid as here https://github.com/sshaoshuai/Pointnet2.PyTorch/blob/master/tools/dataset.py#L165