Marine Biological Image Recognition Based on Deep Convolutional Neural Network

Marine image recognition is a basic research of underwater robots. Because the marine images are weak, variable lighting, low contrast and blurring, marine image recognition is a difficult task. In the traditional marine image recognition methods, feature extraction depends on manual guidance and prior knowledge. And as the amount of data increases, the efficiency is drastically reduced. So, a marine biological identification method based on deep convolutional neural network is proposed. The marine biological image recognition based on deep convolutional neural network (MB-DCNN) has few structural parameters, the accuracy rate reaches 85.9%.