Closed Zumbalamambo closed 7 years ago
Depends on image size. If it's smaller than 96x96, use the DenseNet method to custom build a DenseNet with Cifar mode.
If you are using larger image sizes, look up Keras fine-tuning and substitute VGG for DenseNetImegaNetYYY where Y is the size of the DenseNet.
How do I train my own dataset?