search

experiencor / keras-yolo2

Easy training on custom dataset. Various backends (MobileNet and SqueezeNet) supported. A YOLO demo to detect raccoon run entirely in brower is accessible at https://git.io/vF7vI (not on Windows).
MIT License
1.73k stars 785 forks source link

multi-class classification #331

Open shivaram93 opened 6 years ago

shivaram93 commented 6 years ago

How can train it for custom multi-class classification. does this supports?

NPetsky commented 6 years ago

Yes, I'm doing it also. Just put your labels into the the config.json file like this for example: "labels": ["person", "rider", "car", "truck", "bus", "train", "motorcycle", "bicycle"] and start training