search

tianbaochou / NasUnet

170 stars 45 forks source link

search_cell.py results in pixACC 0.67 & mIoU 0.03 & loss is constant #15

Closed ghost closed 5 years ago

ghost commented 5 years ago

@tianbaochou Thank you for your hard work. I ran python search_cell.py for 18 hours, and used only 300 images from the voc 2012 dataset, the results were mostly pixACC 0.67 & mIoU 0.03

Why is the mIoU very low, pixACC not improving and loss is constant?

The search folder is attached. search-20190516-163658.zip

tensorboard

tianbaochou commented 5 years ago

@tianbaochou Thank you for your hard work. I ran python search_cell.py for 18 hours, and used only 300 images from the voc 2012 dataset, the results were mostly pixACC 0.67 & mIoU 0.03

Why is the mIoU very low, pixACC not improving and loss is constant?

The search folder is attached. search-20190516-163658.zip

tensorboard

Hi, the images you trained is too small, which are natural images. Since we train the architecture from scratch. the model parameters is not sufficient trained. So, both the pixAcc and mIOU are small.