PointPainting
PointPainting copied to clipboard
how to finetune kitti dataset used the bisenet weight? thx
Since the cityscapes dataset has 20 classes. But kitti, it don't have 20 category.
So I have the weight publiced by BiSeNet with the cityscapes, how to finetune in the Kitti dataset.
Really thanks
by fine tuning, this is done by replacing the last layers that depends on the number of detected classes
the code is already ready for that, it has that replaced layers to make it trainable on KITTI dataset classes
Hello, thank you for your excellent research. I would like to ask how you did fine-tuning training on the kitti data set? How to modify it for different input image sizes?