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
787
forks
source link
training on custom data plus taking a pretrained class #425
I want to train the model on say ten custom classes but also want person to be detected too. How can I retain the pretrained class like person and include new classes?
Can just adding person to the labels help without adding any person example in the custom dataset?
I want to train the model on say ten custom classes but also want person to be detected too. How can I retain the pretrained class like person and include new classes?
Can just adding person to the labels help without adding any person example in the custom dataset?