relation

Learning to Compare: Relation Network for Few-Shot Learning using Pytorch

A implementation of the CVPR 2018 paper : Learning to Compare: Relation Network for Few-Shot Learning using pytorch. In the model, somethings, such as learning rates or regression, may differ from the original paper.

I will post the details of the code in Korean on my blog(will be soon.) so if you are interested, please visit!

한글로 논문과 코드에 대해 작성한 글이 있으니 관심있으신 분은 확인해보세요!

🚀How to run

1. Go into prototypical directory

   cd relation

2. Train

   This commend will train the model.

   python train.py

3. Test

   If trained models are exists, you can test the model. Below command will test the entire model in runs/exp_name

   python eval.py

4. Logs

   Train logs, saved model and configuration data were in run/exp_name. Logs are made by tensorboard. So if you want to see more detail about train metrics, write commend on like this.

   tensorboard --logdir=runs

All parameters are present in arguments.py. If you want to adjust the parameters, modify them and run the code.

📈Result

Model	Reference Paper	This Repo
Omniglot 5-w 1-s	99.6±0.2%	99.6%
Omniglot 5-w 5-s	99.8±0.1%	99.7%
Omniglot 20-w 1-s	97.6±0.2%	97.3%
Omniglot 20-w 1-s	99.1±0.1	99.1%
miniImageNet 5-w 1-s	50.44±0.82%	51.2±0.3%
miniImageNet 5-w 5-s	65.32±0.66%	65.6±0.2%

📌Reference

• Learning to Compare: Relation Network for Few-Shot Learning
• floodsung/LearningToCompare_FSL