Marine Organism Detection and Classification from Underwater Vision Based on the Deep CNN Method

Jingzheng Yao,Fenglei Han,Chunhui Wang,Haitao Zhu

doi:10.1155/2020/3937580

Jingzheng Yao, Fenglei Han + Show 2 more

Open Access

https://doi.org/10.1155/2020/3937580

Copy DOI

Abstract

Seabed fishing depends on humans in common, for instance, the sea cucumber, sea urchin, and scallop fishing, which is always a very dangerous task. Considering the underwater complex environment conditions such as low temperature, dim vision, and high pressure, collecting the marine products using underwater robots is commonly regarded as a feasible solution. The key technique of the underwater robot development is to detect and locate the main target from underwater vision. This research is based on the deep convolutional neural network (CNN) to realize the target recognition from underwater vision. The RPN (Region Proposal Network) is used to optimize the feature extraction capability. Deep learning dataset is prepared using an underwater video obtained from a sea cucumber fishing ROV (Remote Operated Vehicle). The inspiration of the network structure and the improvements come from the Faster RCNN and Hypernet method, and for the underwater dataset, the method proposed in this paper shows a good performance of recall and object detection accuracy. The detection runs with a speed of 17 fps on a GPU, which is applicable to be used for real-time processing.

Highlights

Sea urchins, scallops, and other marine organisms live in the bottom of the sea, which are mainly fished by humans, and the fishing process is dangerous
Underwater fishing robots are mostly operated by humans using a remote control device, the fishing speed is slow, and the image information collected by using an underwater camera is generally vague, so it is very difficult to detect the position of sea cucumber and something else by human eyes
E following images are used to testify the method proposed in this paper, the images shown in Figure 10 are filmed by the remote operated vehicle (ROV) in remote location, the objects are tiny, and some of them are overlapped by the other objects; even when the images are vague, the objects can be detected and classified accurately

Summary

Introduction

Sea urchins, scallops, and other marine organisms live in the bottom of the sea, which are mainly fished by humans, and the fishing process is dangerous. Deep learning has more advantages than traditional image processing methods in terms of precision and speed of target detection, and the idea is applied in various fields. Mane and Pujari developed a video detection system based on the Gauss mixed model using a differential method to remove the background information, which can be applied to locate the moving target in the video, and the recognition rate can reach 80% [4]. In the field of image recognition, the development of the algorithm based on convolution neural network is fast. E traditional methods are using contour segmentation and feature extraction to locate the target These methods have already been applied in many fields, the speed and accuracy are far behind the advanced technology in image recognition. These methods have already been applied in many fields, the speed and accuracy are far behind the advanced technology in image recognition. is research intends to propose a new underwater CNN recognition technology to optimize the detection program, which is used to facilitate the underwater ROV to detect and classify marine organism

Methodology Applied in Marine Organism Detection

Improvement and Modifications for Application

Method

Conclusion