xuhao-anhe

_base_ = [ '../../_base_/datasets/mmdet/coco_instance.py', '../../_base_/schedules/mmdet/schedule_1x.py', '../../_base_/mmdet_runtime.py' ] # model settings student = dict( type='mmdet.PointRend', backbone=dict( type='ResNet', depth=18, num_stages=4, out_indices=(0, 1, 2, 3), frozen_stages=1, norm_cfg=dict(type='BN', requires_grad=False), norm_eval=True, style='caffe', # init_cfg=dict( #...

> Sorry for not replying in time because of missing the GitHub notice. Could you provide us with your config for analyzing better? Sorry for the late reply. I haven't...

> @pprp I wanna train voc format dataset using mmdet， I change two places in cwd_cls_head_gfl_r101_fpn_gfl_r50_fpn_1x_coco.py.   > > I have some questions. > > 1. I have two...

Later, I successfully ran through Point_rend with CWD, but at the beginning of the training, all the losses were Nan. What is the reason

2022-06-27 15:25:49,260 - mmdet - INFO - workflow: [('train', 1)], max: 50 epochs 2022-06-27 15:25:49,260 - mmdet - INFO - Checkpoints will be saved to /media/jidong/code/xuhao/mmrazor-master/test by HardDiskBackend. [W TensorIterator.cpp:918]...

> Sorry, we don't know the point rend model very well and can't give very accurate recommendations. At present, there is no very general KD algorithm that can be applied...

Thank you for your contribution. Can other knowledge distillation algorithms under this framework test the instance segmentation algorithm

> You can try [cwd](https://github.com/open-mmlab/mmrazor/tree/master/configs/distill/cwd), and [here](https://github.com/open-mmlab/mmrazor/blob/master/configs/distill/cwd/cwd_cls_head_pspnet_r101_d8_pspnet_r18_d8_512x1024_cityscapes_80k.py) is an example config. Sorry, I can't find a suitable Module in Point_rend for CWD algorithm. Which module of Point_rend do I fill...

Excuse me, can anyone solve this problem

Thank you for your reply， I will check it.