I used the dataset called Cifar10 that contains 60000 32x32 pixel images.
The dataset has 10 classes: airplane, automobile,bird,cat, deer, dog, frog, horse, ship, truck. There are 50000 training and 10000 testing images.
I used Tensorflow Sequential API to create a basic model that has a relatively high accuracy. After I trained a model, I created a prediction on the test set and also used some other downloaded pictures to make predictions.
After I got that right I created my machine learning API using FastAPI. Created a post request for uploading an image and a get request for the prediction.
It was working well so I created a Docker container for it.