AquaProbe: Efficient Adaptation of Frozen Vision Models for Underwater Images

查看原文 / View Original 编辑 / Edit

Computer Science > Computer Vision and Pattern Recognition arXiv:2610.40900 (cs) [Submitted on 29 Oct 2026] Title: AquaProbe: Efficient Adaptation of Frozen Vision Models for Underwater Images Authors: Md Shihabul Islam Shovo, Nishi Kanta Paul, Trung Dong, Xiaomin Lin.

计算机科学 > 计算机视觉与模式识别 arXiv:2610.40900 (cs) [提交于 2026 年 10 月 29 日] 标题：AquaProbe：水下图像冻结视觉模型的高效适配作者：Md Shihabul Islam Shovo, Nishi Kanta Paul, Trung Dong, Xiaomin Lin。

Abstract: Reliable visual perception is essential for safe autonomous underwater exploration. Yet vision models pretrained on terrestrial images become less reliable underwater, where turbidity, backscatter, and wavelength-dependent attenuation alter scene appearance. Existing adaptation methods either spread adapters equally across the encoder depth or select layers using the divergence between two fixed domains.

摘要：可靠的视觉感知对于安全的水下自主探索至关重要。然而，在陆地图像上预训练的视觉模型在水下环境中的可靠性会降低，因为浊度、后向散射和波长相关的衰减会改变场景外观。现有的适配方法要么将适配器均匀分布在编码器深度中，要么利用两个固定域之间的差异来选择层。

In this paper, we introduce AquaProbe for measuring how the location of the domain gap changes with water type and imaging range. AquaProbe separates newly introduced divergence from that accumulated through a frozen vision encoder using paired clean and physically rendered images and allocates a fixed LoRA budget according to the measured profile.

在本文中，我们引入了 AquaProbe，用于测量域间隙的位置如何随水质类型和成像距离而变化。AquaProbe 利用成对的清晰图像和物理渲染图像，将新引入的差异与通过冻结视觉编码器积累的差异分离开来，并根据测量出的分布情况分配固定的 LoRA 预算。

The segmentation mean Intersection-over-Union (mIoU) for the proportional LoRA rank allocation, based on the measured profile, is 70.1; this is a better result than uniform LoRA rank allocation by 1.9 mIoU points while having the same parameter budget of 0.59 million adapter parameters. The three-way router, conditioned on estimated condition clusters, achieves 70.6 mIoU on the standard SUIM split and 63.4 AUDRC under held-out degradation.

基于测量分布的比例 LoRA 秩分配的分割平均交并比 (mIoU) 为 70.1；这比均匀 LoRA 秩分配的结果高出 1.9 个 mIoU 点，同时保持了 59 万个适配器参数的相同参数预算。基于估计条件聚类的三向路由器在标准 SUIM 数据集划分上达到了 70.6 的 mIoU，在留出退化测试中达到了 63.4 的 AUDRC。