A demo for train your own dataset on EfficientNet
Thanks for the >A PyTorch implementation of EfficientNet, I just simply demonstrate how to train your own dataset based on the EfficientNet-Pytorch.
Step 1：Prepare your own classification dataset
Then the data directory should looks like:
-dataset\ -model\ -train\ -1\ -2\ ... -test\ -1\ -2\ ...
Step 2: train and test
(1)You can choose to download the pre-trained model automatically or not by modify the
The pre-trained model is available on >release.
You can download them under the folder
(2)Change some settings to match your dataset.
run efficientnet_sample.py to start train and test
(3)You can get the final results and the best model on