Weakly Supervised Underwater Object Real-time Detection Based on High-resolution Attention Class Activation Mapping and Category Hierarchy

• We've introduced a weakly-supervised strategy for real-time underwater object detection that eases the challenge of creating extensive datasets and reduces dependency on instance labels, boosting the models' generalization and accuracy. • Our method uses a high-resolution attention class activation mapping algorithm and a hierarchical category network to extract more comprehensive object potential regions, enhancing positioning accuracy by minimizing background interference. Additionally, a new parametric spatial loss strategy helps the model focus on context, avoiding local optima. • We've developed a technique to filter noisy pseudo-supervised information during training, improving the quality of training signals and further refining the network's detection capabilities. • An Object Aware Loss function (OA-Loss) has been created, and we propose a multi-loss joint learning approach to supervise the object detection network's training, collectively improving detection performance. Recently, deep learning-based underwater object detection technology has achieved remarkable success. However, the accuracy and completeness of dataset instance annotation are crucial for its success. The quality of underwater images is low, severe objects clustering, and occlusion, acquiring object's annotations demands substantial time and labor costs, while mis annotation and missed annotation can also degrade model performance and limit their application in practical scenarios. To address this issue, this paper presents a novel weakly supervised underwater object real-time detection method, which is divided into two subtasks: weakly supervised object localization and real-time object detection. In the weakly supervised object localization task, we design a novel category hierarchy structure network that integrates the high-resolution attention-class activation mapping algorithm to obtain high-quality object class activation maps, weaken background interference, and obtain more complete object regions. The parameterized spatial loss module is devised to enable the model to escape from local optimal solutions, thus accurately and efficiently obtaining object pseudo-detection annotation boxes. For the real-time object detection task, the single-stage detector YOLOv7 is selected as the basic detection model, and an object perception loss function is designed based on the class activation map to jointly supervise the training process. A method for filtering noisy pseudo-supervision information is proposed to enhance the pseudo-supervision information involved in training. Ablation experiments and multi-method comparison experiments were conducted on the URPC and RUOD datasets, and the results verify the effectiveness of the proposed strategy, and our model exhibits significant advantages in detection performance and detection efficiency compared to current mainstream and advanced models.