Sha-Lab/FEAT
PyTorch implementation of FEAT, a few-shot learning method that adapts embeddings using Transformer set-to-set functions for visual classification.
Not currently ranked — collecting fresh signals.
star history
This repository provides the official PyTorch implementation of a CVPR 2020 paper on few-shot learning. The method adapts pre-trained instance embeddings to target classification tasks using set-to-set functions, with the Transformer architecture proving most effective. The approach is validated on standard few-shot learning benchmarks including MiniImageNet using ResNet-12 backbones, demonstrating improved discriminative task-specific embeddings.
Frequently asked
- What is Sha-Lab/FEAT?
- PyTorch implementation of FEAT, a few-shot learning method that adapts embeddings using Transformer set-to-set functions for visual classification.
- Is FEAT open source?
- Yes — Sha-Lab/FEAT is open source, released under the MIT license.
- What language is FEAT written in?
- Sha-Lab/FEAT is primarily written in Python.
- How popular is FEAT?
- Sha-Lab/FEAT has 434 stars on GitHub.
- Where can I find FEAT?
- Sha-Lab/FEAT is on GitHub at https://github.com/Sha-Lab/FEAT.