Clothes recognition and retrieval on clothing retails.
- tensorflow (2.0.0a0), GPU version can also be used
- opencv (4.1.0.25)
pip install -r requirements.txt
- Run
main.py
to produce final classification result - Run
cloth_detection.py
to produce images with clothes detection bounding boxes.
We trained a Yolo-v3 object detection on DeepFashion2 dataset, pre-trained model weights (tensorflow weights and darknet weights) can be download here.
For the classifier, we use a relatively small dataset consists of only 46 clothes of 2 classes (clothes with stripes and clothes without stripes), the dataset can be download here.