Animal Identification Using CNN

Abstract

Around us there are many kinds of animal with different colors, different sizes, different characteristics, different behavior and different usefulness. In our city’s people mainly city’s children don’t have any or enough idea about our rural animals. Also, our child spent their leisure time by using video games, you-tubing, browsing much unnecessary thing so we realize for them that they should use their time for educating about animal. In my research I have used convolutional neural network or CNN to identify animals. Because CNN is a very popular platform that use for image detection. I have chosen 3 types of animals such as cow, cat and rabbit. Almost 1200 images I have collected to make our dataset. After collected data, I process those and keep 3 different paths. In training path, I keep 60% data and for testing and validation I keep 20%. I have use 4 types of CNN models such as VGG19, RseNet50, Mobile-Net, Inception v3. Then I set and train my dataset into these models and got different types of result. I got best accuracy from Inception v3 model that is 100%. In future I have a plan to add more data and upgrade my system. .......................................................