FGD
FGD copied to clipboard
yzd-v
Focal and Global Knowledge Distillation for Detectors (CVPR 2022)
Can you please provide the code of distillation from Yolox-l to Yolox-m?
Hi, there are some of the loss weights, like https://github.com/yzd-v/FGD/blob/master/mmdet/distillation/losses/fgd.py#L27 How did you tune these loss weights? Any guideline to tune these hyper parameters?
Line 86 in the forward function in mmdet/distillation/losses/fgd.py There are two for loops here, in my test, these two for loops slow down the training seriously, is there any solution?
how about the results of some lightweight backones such as mobbilenetv3。I used resnet50 GFL as teacher and mobilenetv3 GFL as student but got very small improvement in my own datasets....
Traceback (most recent call last): File "tools/train.py", line 15, in from mmdet.apis import set_random_seed, train_detector File "/home/renyu/anzhuang/anaconda3/envs/FGD/lib/python3.7/site-packages/mmdet/apis/__init__.py", line 2, in from .inference import (async_inference_detector, inference_detector, File "/home/renyu/anzhuang/anaconda3/envs/FGD/lib/python3.7/site-packages/mmdet/apis/inference.py", line 7, in...
Add DCN
Hello, Thank you for your insight work! I have a question when I add a DCN structure for the teacher or student, e.g, faster_rcnn_r50_fpn_dconv_c3-c5, cascade_mask_rcnn_x101_32x4d_fpn_dconv_c3-c5, loss_fgd_fpn and total loss will...
mAP为0
我在coco数据集上用yoloxm蒸馏yoloxs，损失曲线也正常，但是第一个epoch，精度为0，我已经按照reademe一步一步改了，之前是在自己数据集上精度为0，我想排查下到底哪里有问题，就在coco上进行了实验，发现在coco上也是同样问题。请问我这可能是哪里的问题那，已经困扰我很多天了。。。。   
When I train my self data, the loss is large, does it normal?The total loss is up 80 in the begining. 
Hi, I had using mask-rcnn-swin-s model with bbox 48.2 mAP and segm 43.2 mAP as teacher,but I got result under baseline,how explain this phenomenon?
