I started learning fast.ai v3 Practical Deep Learning for Coders from youtube by Jeremy Howard. The first lecture was all about image classification and they worked on pet datasets. By gaining the knowledge through that course I implemented it myself.
In the second notebook, I implement the mnist_sample dataset and classify between 3 and 7 which was the homework problem of the course.