SciLT: Long-tailed Image Classification under Scientific Image Domains

TL;DR

This paper introduces SciLT, a framework for scientific long-tailed image classification that leverages multi-level features and dual supervision to improve performance across class distributions.

Contribution

The paper proposes SciLT, a novel method that exploits multi-level representations and dual-supervision learning to enhance scientific long-tailed recognition.

Findings

Fine-tuning foundation models yields limited gains on scientific data.

Penultimate-layer features are crucial for tail class recognition.

SciLT outperforms existing methods across multiple scientific benchmarks.

Abstract

Long-tailed recognition has benefited from foundation models and fine-tuning paradigms, yet existing studies and benchmarks are mainly confined to natural image domains, where pre-training and fine-tuning data share similar distributions. In contrast, scientific images exhibit distinct visual characteristics and supervision signals, raising questions about the effectiveness of fine-tuning foundation models in such settings. In this work, we investigate scientific long-tailed recognition under a purely visual and fine-tuning paradigm. Experiments on three scientific benchmarks show that fine-tuning foundation models yields limited gains, and reveal that penultimate-layer features play an important role, particularly for tail classes. Motivated by these findings, we propose SciLT, a framework that exploits multi-level representations through adaptive feature fusion and dual-supervision learning. By jointly leveraging penultimate- and final-layer features, SciLT achieves balanced performance across head and tail classes. Extensive experiments demonstrate that SciLT consistently outperforms existing methods, establishing a strong and practical baseline for scientific long-tailed recognition and providing valuable guidance for adapting foundation models to scientific data with substantial domain shifts.