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.
Then the data directory should looks like:
-dataset\
-model\
-train\
-1\
-2\
...
-test\
-1\
-2\
...
python efficientnet_sample.py
--data-dir
: (str) Path of /dataset
folder. Default: None
--num-epochs
: (int) Number of epochs for training. Default: 40
--batch-size
: (int) Batch size. Default: 4
--img-size
: (int) Selected size for image to be resized. Default: [1024,1024]
--class-num
: (int) Number of classes in dataset. Default: 3
--weights-loc
: (str) Path of weights to be loaded. If None, pretrained weights will automatically be downloaded & loaded. Default: None
Example: "...//weights.pth//"
--lr
: (float) Learning rate. Default: 0.01
--net-name
: (str) States which efficientnet model will be used. Used for downloading pretrained weights as well.
--resume-epoch
: (int) Defines starting epoch. Default: 0
--momentum
: (float) Sets momentum. Default: 0.9
Example usage: python ".\efficientnet_sample.py" --data-dir "D:\\ml_data\\dataset" --num-epochs 80 --batch-size 4 --img-size 896 --class-num 3 --weights-loc "D:\\ML\\efficientnet-b3-birads.pth" --lr 0.01 --net-name "efficientnet-b3" --resume-epoch 40
The pre-trained model is available on >release.
You can download them under the folder eff_weights
.
(3)You can get the final results and the best model on dataset/model/
.