LoRA Enhanced Contrastive Learning with SAS Vision Transformers

Author1, Author2, Author3, Author4, Author5

LoRAcontrastive learningvision transformersautomatic target recognition

2609.21061

Builder Relevance

70%

1h ago

Abstract

This paper presents a three-stage parameter-efficient framework for automatic target recognition in underwater synthetic aperture sonar using DINOv3 Vision Transformers.

Reality Card

Core Claim

LoRA significantly improves the area under the precision-recall curve (AUPRC) from 0.300 to 0.679 while training only 0.26 percent of weights.

Method / Result

AUPRC increased by 0.379 with LoRA adaptation.

Limitations

The results indicate that additional refinement stages did not significantly improve performance, suggesting limited generalizability of the findings.

Paper to code

Verified implementation resources so builders can test the paper’s claims instead of stopping at the abstract.

No verified implementation link has been attached yet. AIBuzzHub will keep this panel separate from unverified search results.

← Back to all papers