In this project, I used machine learning and data from Kaggle's Dog Breed Identification Competition to help identify different breeds of dogs. This kind of problem is called multi-class image classification, as there are mutliple different breeds of dog.
The model used is one pretrained in the domain of image classification from TensorFlow Hub. It makes predictions with MobileNet-V2, a convolutional neural network that is 53 layers deep and a very effective feature extractor for object detection and segmentation.
Even though there are many ways of building a model in TensorFlow, bone of the best ways if one is getting started is to use the Keras API's Sequential model.
There are a few articles worth reading, but this one by Sumit Saha was definitely my go-to when I was researching Convolutional Neural Networks: https://towardsdatascience.com/a-comprehensive-guide-to-convolutional-neural-networks-the-eli5-way-3bd2b1164a53.
Sumit is a data scientist and machine learning engineer who can do some technical blogging too.