the iou always equals to 0.5 when I use my dataset to train the model

Question

the iou always equals to 0.5 when I use my dataset to train the model

yangyang9706 opened this issue 6 years ago · 6 comments

Thank you for your work. I have the problem that the iou always equals to 0.5 when I use my dataset to train the model. I am looking forward to your answers.

Answer 1 · 2019-04-17T17:28:56.000Z

Did you make any changes to the code?

Answer 2 · 2019-04-18T14:15:16.000Z

I didn't change the code

Answer 3 · 2019-04-18T15:42:24.000Z

What format are your ground truth labels?

Answer 4 · 2019-04-20T07:32:15.000Z

the input image is 250*395, the mask is binarized images whose pixel values are 0 and 1.

Answer 5 · 2019-04-20T19:28:36.000Z

You need to comment out the normalization function in the dataloader, since it divides the groundtruth labels by 255. Your labels are 0 and 1, 1/255 becomes 0, so the network never learns the foreground class, leading to a 0.5 iou.

Answer 6 · 2019-04-21T02:16:14.000Z

thank you so much! I have solved my problem!