Image-Recommendation

This is AI application for image recommendation based on EfficientNet-B7 / Vision Transformer.

How to use

1. Prepare train/test datasets for image recommendation as required below.
2. Create AI model for the recommendation.
   1. Run all notebooks in src directory.
3. Start streamlit application.

Required datasets

• images : {id}.png or {id}.jpg

• label_tran.csv : label for images

  id	like	update
  0000	1	20220909...
  0010	0	20220909...
  0204	0	20220909...
  ...	...	...

• exclude.csv : exclude label

  id
  9587
  1452
  7628
  ...

• redundant.csv : label for not training

  id
  2754
  1752
  8635
  ...