-> The main objective of the project is to build a multi-class image classifier to identify the breed of a dog given an image of of the dog using TensorFlow 2.0 and TensorFlow Hub.
-> The data used (unstructured data) is from Kaggle's Dog Breed Identification Competition
-> Images are pre-processed and tranfer learning is used.
-> Imagenet classification with MobileNet V2 which is a family of neural network architecture for efficient on-device image classification is used for training.
-> Predictions are made on custom images