A Forward and Backward Compatible Framework for Few-shot Class-incremental Pill Recognition

TL;DR

This paper presents a novel few-shot class-incremental learning framework for pill recognition, combining virtual class synthesis and pseudo-feature generation to improve recognition accuracy with limited data and new class updates.

Contribution

It introduces the first framework for few-shot class-incremental pill recognition, integrating forward and backward compatibility strategies with virtual class synthesis and pseudo-feature generation.

Findings

Outperforms existing FSCIL methods on new pill dataset.

Effective virtual class synthesis enhances discriminative features.

Pseudo-feature generation reduces storage needs and maintains old class knowledge.

Abstract

Automatic Pill Recognition (APR) systems are crucial for enhancing hospital efficiency, assisting visually impaired individuals, and preventing cross-infection. However, most existing deep learning-based pill recognition systems can only perform classification on classes with sufficient training data. In practice, the high cost of data annotation and the continuous increase in new pill classes necessitate the development of a few-shot class-incremental pill recognition system. This paper introduces the first few-shot class-incremental pill recognition framework, named Discriminative and Bidirectional Compatible Few-Shot Class-Incremental Learning (DBC-FSCIL). It encompasses forward-compatible and backward-compatible learning components. In forward-compatible learning, we propose an innovative virtual class synthesis strategy and a Center-Triplet (CT) loss to enhance discriminative feature learning. These virtual classes serve as placeholders in the feature space for future class updates, providing diverse semantic knowledge for model training. For backward-compatible learning, we develop a strategy to synthesize reliable pseudo-features of old classes using uncertainty quantification, facilitating Data Replay (DR) and Knowledge Distillation (KD). This approach allows for the flexible synthesis of features and effectively reduces additional storage requirements for samples and models. Additionally, we construct a new pill image dataset for FSCIL and assess various mainstream FSCIL methods, establishing new benchmarks. Our experimental results demonstrate that our framework surpasses existing State-of-the-art (SOTA) methods. The code is available at https://github.com/zhang-jinghua/DBC-FSCIL.