Convolutional Neural Networks (CNN) based Marine Species Identification

There are more than 5600 species of fish in the world. Studies show that crowds can be larger and many people are less knowledgeable about the characteristics of marine life. This project is to raise public awareness and generate knowledge about Aquaculture. In fisheries education, it is important to minimize human error in the process of fish observation and analysis, where an automatic system recognition is required. Image processing is also known as the process of taking image as an input, analyzing it and generating a category or probability of a particular class of output type. Fish image recognition research is an important area, especially in marine biology and aquaculture. Recently, the identification of different classes of animals based on computer vision and deep learning has been widely used by analysts. There is one of the ultimate powerful forms of deep learning architecture for conceptual classification based on visual features. It is reported that there are many kinds of clean water fish in world level. Most of the existing work has been developed for the identification of fewer fish species and used for specific user groups. This detection system is used to identify and classify images of fish species found in clean water as well as in sea water.