Improved YOLO v5 with principal component analysis for river object detection

Abstract

River pollution not only damages the environment but also hinders the overall development of cities. With the development of the economy, river management has gradually become more important. An image processing method based on principal component analysis is proposed, which preprocesses the data through principal component analysis algorithm, and standardizes the data. Then, a YOLOv5 algorithm detection model for river scenes is proposed. Attention mechanism is introduced to detect floating objects in the river through the YOLOv5 algorithm model. The experimental results showed that when the training set size was 800, the intersection over union ratios of factor analysis, independent component analysis, and principal component analysis models were 0.85, 0.92, and 0.98, respectively. The structural information loss was 0.12, 0.08, and 0.04, respectively. The fuzziness was 0.21, 0.18, and 0.12, respectively, and the signal-to-noise ratios were 0.85, 0.92, and 0.98, respectively. The research results indicate that the proposed improved model has excellent performance, which can promote the improvement of river and lake conditions and provide technical support for long-term effective river and lake supervision, providing new ideas for assisting river inspection and supervision.