Abstract:
Ornamental fish are the most important fishing commodity in the world in terms of unit weight,
with a global export volume of about 350 million dollars. Ornamental fish are grown using the
same simple technology and nutrients as food fish, but ornamental fish farming has not historically
been considered an aquaculture practice. This is due in part to the fact that literature on different
facets of ornamental fish culture is often conducted in books and specialized magazines rather than
peer-reviewed publications, or it is held proprietary. Technology has the potential to make a
significant contribution to all facets of our lives. Deep learning algorithms that use a particular
kind of neural network called a convolutional neural network (CNN) to make sense of images are
at the heart of today's computer vision technologies. In deep learning, I will use the convolutional
neural network (CNN) to achieve state-of-the-art precision in a variety of classification problems,
such as image data, CIFAR-100, CIFAR-10, and MINIST data sets. In this paper, I propose a novel
system that uses Convolutional neural networks to classify different types of ornament fish
detections, automatic self-ruling decision making, and predictive models (CNN). While there has
been a number of studies on fish picture detections in image classification problems in the past,
our associated tropic ornament fish type detection issue has just a few works on various data sets
and different models with low accuracy. I retrained the final layer of the CNN architecture,
VGG16, Inception V3 for classification strategy, for solid architecture. Predicting between five
groups (goldfish, Arowana fish, betta fish, angelfish, rainbow shark). I suggested a 95% average
accuracy that can be used for a variety of uses, such as purchasing fish, classification and assisting
in the management of a large aquarium.
