Underwater object detection using Invert Multi-Class Adaboost with deep learning

TLDR

Deep learning has achieved promising performance in standard object detection, yet underwater images pose challenges such as small, blurry objects and heterogeneous noise that limit existing methods. This study proposes the Sample‑Weighted hyPEr Network (SWIPENet) to improve small‑object detection in underwater settings. SWIPENet employs high‑resolution hyper feature maps and a sample‑weighted loss optimized by an Invert Multi‑Class Adaboost algorithm to mitigate noise effects. Experiments on the URPC2017 and URPC2018 datasets show that SWIPENet+IMA outperforms several state‑of‑the‑art underwater object detectors.

Abstract

In recent years, deep learning based methods have achieved promising performance in standard object detection. However, these methods lack sufficient capabilities to handle underwater object detection due to these challenges: (1) Objects in real applications are usually small and their images are blurry, and (2) images in the underwater datasets and real applications accompany heterogeneous noise. To address these two problems, we first propose a novel neural network architecture, namely Sample-WeIghted hyPEr Network (SWIPENet), for small object detection. SWIPENet consists of high resolution and semantic-rich Hyper Feature Maps which can significantly improve small object detection accuracy. In addition, we propose a novel sample-weighted loss function which can model sample weights for SWIPENet, which uses a novel sample re-weighting algorithm, namely Invert Multi-Class Adaboost (IMA), to reduce the influence of noise on the proposed SWIPENet. Experiments on two underwater robot picking contest datasets URPC2017 and URPC2018 show that the proposed SWIPENet+IMA framework achieves better performance in detection accuracy against several state-of-the-art object detection approaches.