This project use Retina Net to do SVHN digit detection.

SVHN is obtained from house numbers in Google Street View images. It contain 10 classes, 1 for each digit. Digit '1' has label 1, '9' has label 9 and '0' has label 10 instead of 0.

torch version 1.7.0

1. Download dataset from official site http://ufldl.stanford.edu/housenumbers/
2. Clone this repo
3. Install Python packages

pip install pandas
pip install pycocotools
pip install opencv-python
pip install requests

Run train.py by following command

python train.py --dataset csv --csv_train <path/to/train_annots.csv>  --csv_classes <path/to/train/class_list.csv>  --csv_val <path/to/val_annots.csv>

pre-trained model is available at: https://drive.google.com/file/d/1-Szeyz9LA5X_-MvbuYriWovQPEYyrKVI/view?usp=sharing

To visualize image, use:

python inference.py --dataset csv --csv_classes <path/to/train/class_list.csv>  --csv_val <path/to/val_annots.csv> --model <path/to/model.pt>

1. https://github.com/pavitrakumar78/Street-View-House-Numbers-SVHN-Detection-and-Classification-using-CNN/blob/master/train_digit_detection.py
2. https://github.com/AlessandroSaviolo/RetinaNET-Object-Detector/tree/master/src
3. https://github.com/yhenon/pytorch-retinanet