Giters

oist / Usiigaci

Usiigaci: stain-free cell tracking in phase contrast microscopy enabled by supervised machine learning

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

how to improve the accuracy of model

w1n2 opened this issue · comments

commented

I use own dataset(DIC microscope images) to runing train.py. I don't modify other parameters. And I found the loss cann't be reduced after about 0.9. Can you give me some help to reduce the loss. My DIC microscope images have lower contrast.