Few-Shot Learning for Skin Analysis

Model building, experiments, references and source-code for the research work on skin image analysis using few-shot learning.

Cite Us

Link to the Research Paper Preprint

If you find our work useful in your research, cite us:

@article{https://doi.org/10.48550/arxiv.2210.16954,
  doi = {10.48550/ARXIV.2210.16954},  
  url = {https://arxiv.org/abs/2210.16954},  
  author = {Desingu, Karthik and P., Mirunalini and Chandrabose, Aravindan},  
  keywords = {Computer Vision and Pattern Recognition (cs.CV), Artificial Intelligence (cs.AI), Machine Learning (cs.LG), FOS: Computer and information sciences, FOS: Computer and information sciences},  
  title = {Few-Shot Classification of Skin Lesions from Dermoscopic Images by Meta-Learning Representative Embeddings},  
  publisher = {arXiv},  
  year = {2022},  
  copyright = {Creative Commons Attribution 4.0 International}
}

Proposed Embedding-Network & Base-Learner Approach for Meta Learning

This work proposes a two-stage approach: a baseline supervised learner on the meta-training set that allows a network to learn highly representative and generalizable feature embeddings for images, that are readily transferable to new few-shot learning tasks; positing that a representative feature embedding can be more effective than complex meta-learning algorithms, a simple linear classifiers is trained atop these representations.

Key References

[CVPR-2020] Meta-DermDiagnosis Few-Shot Skin Disease Identification using Meta-Learning.pdf

Proposes the use of meta-learning techniques for efficient model adaptation for extremely low-data scenarios
Applies Group equivariant convolutions (G-convolutions) in place of the normal spatial convolution filters
Two network implementations:
- Reptile: Gradient-based meta-learning
- Prototypical networks using Euclidean Distance
Evaluated on ISIC 2018, Derm7pt and SD-198 datasets
Outperforms DAML on ISIC 2018
Implementation Code NOT available

[CVPR-2018] Learning to Compare Relation Network for Few-Shot Learning

The paper that proposed Relation Networks for Few-Shot Learning

[NeurIPS-2017] Prototypical Networks for Few-shot Learning

The paper that proposed Protoypical Networks for Few-Shot Learning

[Elsevier-PR-2020] Temperature network for few-shot learning with distribution-aware large-margin metric

An improvement of Prototypical Networks, by generating query-specific prototypes and thus results in local and distribution-aware metric
Sets different temperature for different categories to penalize query samples that are not close enough to their belonging categories.
Code available - PyTorch

autocrat-vst / SkinAnalysisLearning