ROIMIX: Proposal-Fusion Among Multiple Images for Underwater Object Detection

TLDR

Underwater object detection remains underexplored due to challenges such as color shift, low contrast, sediment-induced blurring, and the frequent close proximity of marine creatures, despite the strong performance of generic detection algorithms on terrestrial datasets. The study aims to improve underwater object detection by developing augmentation policies that simulate overlapping, occluded, and blurred objects to enhance model generalization. The authors introduce RoIMix, an augmentation technique that mixes proposals from multiple images to generate enhanced training samples, contrasting with prior single-image methods. Experiments demonstrate that RoIMix boosts region‑based detector performance on both Pascal VOC and URPC datasets.

Abstract

Generic object detection algorithms have proven their excellent performance in recent years. However, object detection on underwater datasets is still less explored. In contrast to generic datasets, underwater images usually have color shift and low contrast; sediment would cause blurring in underwater images. In addition, underwater creatures often appear closely to each other on images due to their living habits. To address these issues, our work investigates augmentation policies to simulate overlapping, occluded and blurred objects, and we construct a model capable of achieving better generalization. We propose an augmentation method called RoIMix, which characterizes interactions among images. Proposals extracted from different images are mixed together. Previous data augmentation methods operate on a single image while we apply RoIMix to multiple images to create enhanced samples as training data. Experiments show that our proposed method improves the performance of region-based object detectors on both Pascal VOC and URPC datasets.