EfficientVPR：基于场景感知提示词微调与自适应特征增强的高效视觉位置识别
EfficientVPR: Toward Efficient Visual Place Recognition via Scene-Aware Prompt Tuning and Adaptive Feature Enhancement
用场景感知的视觉提示词微调加实例相关的局部关键特征增强，在 DINOv2-small 上以低维特征实现 73 倍加速，为视觉检索类任务提供了一条「不靠堆模型容量、不靠重排序」的效率路线，可直接迁移到图像检索与重定位的特征提取环节。
原文依据 a scene-aware visual prompt tuning method which adapts pretrained features with less parameters while dynamically adjusting to sample-specific characteristics
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论视觉位置识别（VPR）在应对剧烈环境变化的同时还需满足实际应用的计算约束，面临严峻挑战。现有方法大多通过扩大模型容量或引入计算密集的重排序阶段来应对这些挑战，造成了显著的效率瓶颈。为克服这一局限，我们提出 EfficientVPR，一个轻量级单阶段框架，通过两项关键创新实现速度与精度的权衡：i）场景感知的视觉提示词微调方法，以更少参数适配预训练特征，同时动态适应样本特有的特性；ii）实例相关的关键局部特征增强模块，进一步强化判别性区域。在 Pitts250k、MSLS、Eynsham、AmsterTime 和 SVOX 上的全面评测表明，本方法在 DINOv2-small 模型上建立了新的 SOTA，优于所有同规模竞争者，并在特征维度降低 60% 的情况下实现 73 倍加速，同时与基于 DINOv2-large 的两阶段 SOTA 方法保持有竞争力的性能（平均 R@1 差距在 2.5% 以内）。我们的代码将发布于 https://github.com/WiniTang/EfficientVPR。
英文原文摘要 1245 字符
Visual place recognition (VPR) faces critical challenges in handling extreme environmental variations while meeting the computational constraints of practical applications. Current methods predominantly address these challenges by either scaling up model capacity or employing computationally intensive reranking stages, creating a significant efficiency bottleneck. To overcome this limitation, we propose EfficientVPR, a lightweight one-stage framework that achieves speed-accuracy trade-offs through two key innovations: i) a scene-aware visual prompt tuning method which adapts pretrained features with less parameters while dynamically adjusting to sample-specific characteristics, and ii) an instance-dependent key local feature enhancement module that further reinforces discriminative regions. Comprehensive evaluations on Pitts250k, MSLS, Eynsham, AmsterTime and SVOX demonstrate that our method establishes a new SOTA for DINOv2-small models by outperforming all same-scale competitors, and delivers a 73x speedup with 60% lower-dimensional features while maintaining competitive (within 2.5% average R@1 gap) against the SOTA DINOv2-large-based two-stage method. Our code will be available at https://github.com/WiniTang/EfficientVPR.