Multi-scale underwater fish school detection method based on attention module

Abstract

The invention discloses a multi-scale underwater fish school detection method based on an attention module, and belongs to the technical field of marine fish resource detection. Firstly, underwater fish school images are obtained, fish school images are selected from the underwater fish school images, enhancement preprocessing is conducted on the underwater images according to the characteristics that the underwater environment is dark, the background is complex and the fish school images are small and dense, based on an optimized YOLO V4 detection framework, an ECA attention mechanism is added to enable a network to focus on fish body feature learning, background interference is overcome, and the fish school images are obtained. PANet connection is improved, multi-scale information is added, feature extraction is enhanced, and the detection precision is improved. According to the method, automatic detection of underwater fish schools is realized; scientific theory and technical support are provided for automatic fish detection, scientific fishing is specified, ocean resources can be better protected, and sustainable and healthy development of an ocean system is promoted; and a reference scheme can be provided for detection of other marine organisms.

Description

technical field [0001] The invention belongs to the technical field of detection of marine fish resources, and in particular relates to a multi-scale underwater fish detection method based on an attention module. Background technique [0002] The ocean accounts for about 71% of the total area of ​​the earth, and there are abundant marine biological resources such as fish and minerals, which can bring huge economic benefits to human production and life. Among them, fish resources are important marine resources, and the detection and identification of fish is of great significance for exploring the law of marine fish activities, monitoring the water environment, and conducting research on fishery resources. However, with the improvement of the level of science and technology and the continuous expansion of market demand, the intensity of fishing has increased rapidly and the scope of fishing seas has expanded significantly. In recent years, predatory fishing has appeared. mar...

AI Technical Summary

Problems solved by technology

[0011] To sum up, at present, there are still defects in avoiding overfishing and protecting marine biological resources due to the variety of fish in the ocean, the low quality of underwater imaging, and the difficulty of fish detection.

Images