Demo :
download weights and put at folder detection
https://drive.google.com/file/d/1YLC88RKG-2IYnhqW7uWRHIgbi04JWk4p/view?usp=sharing
python3 test.py ( with license plate)
you can edit config at detection/config.py and detection/data/config.py ( mostly image_target )
Train: (go to folder detection)
Dataset:
make data by labelme tool
example in data/dataset
you can edit custom_dataset that has format similar detection/data/custom_data.py
Run:
1) Edit config in train.py ( have 2 backbone mobilenet0.25 and resnet50)
2) python3 train.py
References:
https://github.com/wkentaro/labelme
https://github.com/yhenon/pytorch-retinanet
Goodluck for you!
if you have any problems please contact me (https://www.facebook.com/hoangquoc.hao)