MLP-Mixer

An Implementation for paper MLP-Mixer: An all-MLP Architecture for Vision using PyTorch

How to use

To clone and run this application, you'll need Git installed on your computer. From your command line:

Clone this repository

$ git clone https://github.com/hungsvdut2k2/MLP-Mixer.git

Go into the repository

$ cd MLP-Mixer

Install packages

$ pip install -r requirements.txt

Set up your dataset

Structure of these folders.

train/
...class_a/
......a_image_1.jpg
......a_image_2.jpg
...class_b/
......b_image_1.jpg
......b_image_2.jpg
...class_c/
......c_image_1.jpg
......c_image_2.jpg

validation/
...class_a/
......a_image_1.jpg
......a_image_2.jpg
...class_b/
......b_image_1.jpg
......b_image_2.jpg
...class_c/
......c_image_1.jpg
......c_image_2.jpg

Train your model by running this command line

$ python train.py --epochs ${epochs} --num-classes ${num_classes}