研究目的
Investigating the effectiveness of a novel underwater object detector that introduces multi-scale features and complementary context information for better classification and location ability in complex underwater environments.
研究成果
The proposed underwater object detector, incorporating multi-scale features and complementary context information, achieves superior performance on the challenging NSFC-dataset, outperforming four typical state-of-the-art detectors. The model is efficient, achieving 24 FPS on a portable device, making it suitable for real-time applications.
研究不足
The study focuses on underwater object detection in complex environments, which may not generalize well to other contexts. The model's performance is evaluated on a specific dataset, limiting its applicability to other underwater scenarios without further validation.
