Loss and val_loss improvement

david8862 / keras-YOLOv3-model-set

end-to-end YOLOv4/v3/v2 object detection pipeline, implemented on tf.keras with different technologies

MIT License

640 stars 222 forks source link

Open dikshaaswal opened 4 years ago

dikshaaswal commented 4 years ago

Thanks for the wonderful code. I am trying to use diou loss, but after ep010-loss3.02-val-loss3.123, there is no improvement. How to fix this.

david8862 commented 4 years ago

@dikshaaswal not sure about your training detail. Have you ever tried to reduce the learning rate?

dikshaaswal commented 4 years ago

The training details are same as your's given in the code except using DIoU loss.

rootadminWalker commented 4 years ago

@dikshaaswal Have you found the solution for loss improve? cuz I'm experiencing the same problem right now