search

akirasosa / mobile-semantic-segmentation

Real-Time Semantic Segmentation in Mobile device
MIT License
715 stars 135 forks source link

Any pretrained model available? #17

Closed yangchen9 closed 6 years ago

yangchen9 commented 6 years ago

could you give a pretrained model for finetune?

mayurbhangale commented 6 years ago

Releasing pretrained models- https://drive.google.com/open?id=1DaZPwr4_O5xQ_QRGyzk6_5egk1nZuHhP

ldenoue commented 6 years ago

Thanks for posting. I notice that it only segments the hair. See my own portrait image which I ran through the predictor: laurent_denoue output

mayurbhangale commented 6 years ago

From readme:

The goal of this project is to detect hair segments with reasonable accuracy and speed in mobile device. Currently, it achieves 0.89 IoU.

ldenoue commented 6 years ago

Thanks, so I'll need to change the training masks to get both classes. Got it, thanks again.

nemanjaq commented 6 years ago

Hello @ldenoue ,

can you tell me what code did you use to run single picture and get that result. I'm new to ML.

Thanks

akirasosa commented 6 years ago

I have rewrite code using Pytorch. It's much simpler than before. And uploaded pre trained weight. https://github.com/akirasosa/mobile-semantic-segmentation#pretrained-model