set weight for positive class for segmentation

qubvel / segmentation_models

Segmentation models with pretrained backbones. Keras and TensorFlow Keras.

MIT License

4.73k stars 1.03k forks source link

Open nyk2001 opened 3 years ago

nyk2001 commented 3 years ago

Hi,

I am using UNET for segmentation. I want to increase the weight of my positive class.

Because number of pixels for positive class is very low compared to non-positive pixels.

Is there a way I can do it ?

Thanks