SciLT: Long-Tailed Classification in Scientific Image Domains

2026-04-04

AI Summary

This work addresses the challenge of long-tailed classification in scientific images, where significant domain shifts from natural images limit the effectiveness of standard foundation model fine-tuning approaches. The study presents the first systematic analysis of the limitations of foundation models in this setting and introduces SciLT, a parameter-efficient fine-tuning framework that adaptively fuses multi-level features and employs dual supervision to jointly optimize both the final layer and the penultimate layer—features particularly critical for tail-class recognition. By explicitly balancing performance across head and tail classes, SciLT achieves substantial gains over existing methods on three scientific image benchmarks, establishing a strong and practical new baseline for long-tailed recognition in scientific domains.

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 parameter-efficient fine-tuning (PEFT) 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.

Problem

• long-tailed recognition
• scientific images
• foundation models
• domain shift
• fine-tuning

Innovation

• long-tailed recognition
• scientific images
• parameter-efficient fine-tuning
• multi-level feature fusion
• dual-supervision learning