Computer Vision: Identifying Dog Breeds with CNN

Victoria Yuanyuan Chang

Abstract:

The current project aims at identifying dog breeds from images. The model has various potential applications, for example, identifying lost pets in security footage. The data used in this project is extracted from Kaggle. The data consists of images of various dogs and a .csv file that maps the breed name to the images. The main method of the project is CNN. I first preprocessed the data using methods such as data augmentation, then split the images into training and testing data using tts. Finally, I built a CNN model and fitted it to the data. The model is was validated by accuracy.

Video Presentation:

https://drive.google.com/file/d/1ccrr2_R9rKG2dzppqDI2mKjs-NcMGbkz/view?usp=sharing