Skip to content

Realtime segmentation with ENet, the fast and accurate segmentation net.

Notifications You must be signed in to change notification settings

lucasjinreal/pt_enet

Repository files navigation

Enet

This is a realtime segmentation net with almost 22 fps on GTX1080 ti, and the model size is very small with only 28M.

This repo contains the inference demo with already trained weights for production. the weights trained on cityscapes and camvid. the result like below:

Runing

To run the demo, simply:

python3 demo.py

this demo requires alfred-py, you should install first with:

sudo pip3 install alfred-py

the alfred-py is a helpfull utilize library for deep learning: http://github.com/jinfagang/alfred

Training

If you want training with your own dataset or fine-tune on cityscapes, you can get the full version codes from http://strangeai.pro . contact me to get the full codes, you can find me on that website.

you can access now: http://strangeai.pro/aicodes_detail.html?id=24

About

Realtime segmentation with ENet, the fast and accurate segmentation net.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages