Dog Breed Identification using ResNet Model
Yarram Abhilash Reddy, Yambaku Suneel Kumar, M. Sankari, Suja Cherukullapurath Mana · International Journal on Recent and Innovation Trends in Computing and Communication · 2023
As dogs are domestic animals due to the many numbers of dog breeds available around the world. It’s hard to find out the exact dog breed name for a common person. There are many techniques available to identify dog breed. But the proposed work introduced the new technique called RESNET which is the part of CNN to classify dog. RESNET is used to identify images. It helps to perform different tasks on larger datasets. Identification of different dogs is one of the important applications of Convolutional Neural networks. Since the identification of dog breeds is very difficult because they spread in a large number and it makes very hard for a person to identify or classify dogs. With the help of Keras and TensorFlow, a dataset is created, tested, and trained for the detection of dog breeds by using RESNET. Around 120 different dog breeds are present in the dataset which consist of 20600 images of dogs. From this paper, load these images and convert them into a NumPy array and normalize them. Then,100 epochs were used with a batch size of 128 to achieve the best accuracy. The model is saved for further process to create a web application to identify the dog.