Overview / Usage

This project is based on Deep Learning neural network library PyTorch. Model is trained on 133 different breed types. I used Azure NVIDIA Tesla K80 GPU for training model. I was able to achieve 83.16% overall accuracy. I used VS extenstion named PythonNet to pass image path to python script which in turn supplies image to trained neural network. Windows form based application is used to select image file and display predicted breed with Dog image.

This project later can be extended to identify various animals.

Methodology / Approach

Dataset contains different 133 types of dog breeds. They are separated into 3 different categories such as training, validating and testing dataset. Then classifier is written using PyTorch framework.

Technologies Used

• PyTorch
• .NET
• NVIDIA tesla k80 GPU

Repository

https://github.com/vp393001/Dog-Breed-Classification-Using-PyTorch