https://github.com/ajaichemmanam/centerseg
This project uses Centernet and Conditional Convolutions for Instance Segmentation
https://github.com/ajaichemmanam/centerseg
anchor-free centernet conditional-convolutions object-detection object-segmentation
Last synced: 6 months ago
JSON representation
This project uses Centernet and Conditional Convolutions for Instance Segmentation
- Host: GitHub
- URL: https://github.com/ajaichemmanam/centerseg
- Owner: ajaichemmanam
- License: mit
- Created: 2020-05-27T07:38:29.000Z (over 5 years ago)
- Default Branch: master
- Last Pushed: 2020-09-17T04:20:42.000Z (about 5 years ago)
- Last Synced: 2025-04-11T03:52:24.384Z (6 months ago)
- Topics: anchor-free, centernet, conditional-convolutions, object-detection, object-segmentation
- Language: Python
- Homepage:
- Size: 6.1 MB
- Stars: 12
- Watchers: 2
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# CenterSeg
This repo uses Centernet and Conditional Convolutions for Instance Segmentation
> [**Objects as Points**](http://arxiv.org/abs/1904.07850),
> [**CondInst: Conditional Convolutions for Instance Segmentation**](https://arxiv.org/abs/2003.05664)
## Result
These results are taken for CenterSeg model trained for 101 epochs
| type | AP | AP50 | AP75 | APs | APm | APl |
| ---- | ----- | --------------- | --------------- | -------------- | -------------- | -------------- |
| box | 0.278 | 0.430 | 0.297 | 0.129 | 0.305 | 0.382 |
| mask | 0.226 | 0.387 | 0.227 | 0.078 | 0.253 | 0.340 |
| type | AR | AR50 | AR75 | ARs | ARm | ARl |
| ---- | ----- | --------------- | --------------- | -------------- | -------------- | -------------- |
| box | 0.275 | 0.455 | 0.480 | 0.265 | 0.510 | 0.674 |
| mask | 0.235 | 0.369 | 0.385 | 0.170 | 0.418 | 0.585 |
CenterPoseSeg model not trained yet
## Installation
This repo supports both CPU and GPU Training and Inference.
```
git clone --recurse-submodules https://github.com/ajaichemmanam/CenterSeg.git
pip3 install -r requirements.txt
```
Compile DCN
```
cd src/lib/models/networks/DCNv2/
python3 setup.py build develop
```
Compile NMS
```
cd src/lib/external
python3 setup.py build_ext --inplace
```
## Pre-Trained Models
Pre-Release : [Google Drive](https://drive.google.com/drive/folders/1Uw0ucRLpyyHT0pGW2N0o5BcYdSfdNYyC?usp=sharing)
Download the most recent model (model_last_e101.pth), copy to exp/ctseg/coco_dla_1x/
Rename as model_last.pth
```
python3 demo.py ctseg --exp_id coco_dla_1x --keep_res --resume --demo ../data/coco/val2017
```
Note: Model is not completely trained (101 Epochs only). Will update later.
#### Training
###### For GPU
```
python3 main.py ctseg --exp_id coco_dla_1x --batch_size 10 --master_batch 5 --lr 1.25e-4 --gpus 0 --num_workers 4
```
###### FOR CPU
```
python3 main.py ctseg --exp_id coco_dla_1x --batch_size 2 --master_batch -1 --lr 1.25e-4 --gpus -1 --num_workers 4
```
#### Testing
```
python3 test.py ctseg --exp_id coco_dla_1x --keep_res --resume
```
## License
CenterSeg is released under the MIT License (refer to the LICENSE file for details).
This repo contains code borrowed from multiple sources. Please see their respective licenses.
## Credits
https://github.com/xingyizhou
https://github.com/Epiphqny
https://github.com/CaoWGG