Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/ooooverflow/BiSeNet
BiSeNet based on pytorch
https://github.com/ooooverflow/BiSeNet
Last synced: 25 days ago
JSON representation
BiSeNet based on pytorch
- Host: GitHub
- URL: https://github.com/ooooverflow/BiSeNet
- Owner: ooooverflow
- Created: 2018-11-20T06:46:45.000Z (about 6 years ago)
- Default Branch: master
- Last Pushed: 2020-03-13T08:40:57.000Z (almost 5 years ago)
- Last Synced: 2024-08-03T21:02:26.805Z (4 months ago)
- Language: Python
- Size: 1.14 MB
- Stars: 391
- Watchers: 5
- Forks: 77
- Open Issues: 22
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
- awesome-AutoML-and-Lightweight-Models - ooooverflow/BiSeNet
README
# BiSeNet
BiSeNet based on pytorch 0.4.1 and python 3.6
## Dataset
Download CamVid dataset from [Google Drive](https://drive.google.com/file/d/1KRRME_NtRG-iWOyLAb7gE-eA8fTeyzUR/view) or [Baidu Yun](https://pan.baidu.com/s/16k_hSycb2wxmN3IJPpbYig)(6xw4).
## Pretrained model
Download `best_dice_loss_miou_0.655.pth` in [Google Drive](https://drive.google.com/open?id=1ulUgHwFct-vFwGCAfJ4Oa9DBlNDzm5r4) or in [Baidu Yun](https://pan.baidu.com/s/1wHyO0fJhf8j93O90Cn27tA)(6y3e) and put it in `./checkpoints`
## Demo
```
python demo.py
```
### Result
Original | GT |Predict
:-:|:-:|:-:
|
|
## Train
```
python train.py
```
Use **tensorboard** to see the real-time loss and accuracy
#### loss on train
#### pixel precision on val
#### miou on val
## Test
```
python test.py
```
### Result
class|Bicyclist|Building|Car|Pole|Fence|Pedestrian|Road|Sidewalk|SignSymbol|Sky|Tree|miou
:-:|:-:|:-:|:-:|:-:|:-:|:-:|:-:|:-:|:-:|:-:|:-:|:-:
iou | 0.61 | 0.80 |0.86|0.35|0.37|0.59|0.88|0.81|0.28|0.91|0.73|0.655
This time I train the model with **dice loss** and get better result than **cross entropy loss**. I did not use lots special training strategy, you can get much better result than this repo if using task-specific strategy.
This repo is mainly for proving the effeciveness of the model.
I also tried some simplified version of bisenet but it seems does not preform very well in CamVid dataset.
### Speed
Method|640×320|1280×720|1920×1080
:-:|:-:|:-:|:-:
Paper|129.4|47.9|23
This Repo|126.8|53.7|23.6
This shows the speed comparison between paper and my implementation.
1. The number in first row means input image resolution.
2. The number in second and third row means FPS.
3. The result is based on resnet-18.
## Future work
* Finish real-time segmentation with camera or pre-load video
## Reference
* [Semantic-Segmentation-Suite](https://github.com/GeorgeSeif/Semantic-Segmentation-Suite/tree/master)
* [BiSeNet-paper](https://arxiv.org/pdf/1808.00897v1.pdf)