search

WongKinYiu / yolov7

Implementation of paper - YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors
GNU General Public License v3.0
13.23k stars 4.18k forks source link

train loss decrease but val loss increase #120

Open Plwy opened 2 years ago

Plwy commented 2 years ago

Same dataset trained on yolov5 and yolov7.the dataset works on yolov5. but when i training yolov7, the training loss decreases and the accuracy increase , but the val loss increase . what could it mean? is it still on the right track ?should i stop the learning ?

Any suggestion would be much appreciated, thank you!!!!

The figures blow shows the output from tensorboard yolov7: image-20220713104115590

yolov5 : image-20220713104240543

huangxiaosa commented 2 years ago

same question,work fine on yolov5, while val loss increase and map is much lower on yolov7

截屏2022-07-18 上午10 14 10
tonyostum2022 commented 2 years ago

I have the same problem just like you, is this caused by overfitting? Help!

xmayyy commented 2 years ago

me too

jjeremy40 commented 2 years ago

me too...

SherrySwift commented 2 years ago

me too, can anybody give some advice?