The class number of pretrained model should be equal to my custom dataset class number?

qqwweee / keras-yolo3

A Keras implementation of YOLOv3 (Tensorflow backend)

MIT License

7.14k stars 3.44k forks source link

The class number of pretrained model should be equal to my custom dataset class number? #706

Open tjmannn opened 4 years ago

tjmannn commented 4 years ago

It seems the class number is used to build the model, so different class number between pretrained model and custom dataset can cause error? I saw in train.py, the create_model function use "skip_mismatch", is this the way to solve the metioned problem?