training tip

[freaad]

Do you have any tip for small object trainning.
I am dealing the small binary segmentation but it doesn';t have good result.
Do you have any tip.

[lessw2020]

A generic tip is to use focal loss for training with small objects. Not sure if it will work here but it will certainly help in general.

[freaad]

I have already try but not a good result....

[lessw2020]

I'm testing some other loss functions tomorrow and will let you know if any progress. Boundary loss penalty looks really good.

[freaad]

Thank you for answering. I agree the boundary loss penalty is good idea. However my problem is quite imblance data. Background is 95 % the target object is only 5 percent in the image. My problem is only binary segmentation but I cannot get the 90% iou score. I usually get the 83%. I try this network but still no luck. I am trying to use the synchronize batch normalization. Wish me good luck.