Every day, millions of pieces of mail pass through postal services around the world. The U.S. Postal Service alone processes 493.4 million pieces of mail each day, and every piece of mail needs to be individually handled and examined in order to determine the final destination of the specific letter or package by postal code. Needless to say, if done manually, this process requires a great amount of manpower.
Using deep learning for the classifying of handwritten numbers in these postal codes would allow for the process to be automated and the time spent on examining millions of letters vastly reduced.
Click here for the full-text version of the tutorial.
In our tutorial “Deploy an operational AI model,” you are guided through the task of learning how to construct a deep learning model using the Peltarion Platform for the purpose of classifying images with handwritten digits. You will also be taught how to deploy your trained experiment - so you can start using your AI application straight away!
We use the MNIST database of handwritten numbers, a large, well-known dataset of handwritten digits, commonly used for training various image-processing systems. It consists of small, 28x28-pixel grayscale images of handwritten numbers annotated with labels indicating the correct numbers. The dataset consists of a training set of 60,000 examples of images of handwritten numbers and a test set of an additional 10,000 images.
What you will learn
/ Building and training a model for solving a classification problem - meaning, your experiment is about predicting a label, in this case, what number an image depicts.
/ Deploying a trained experiment, so you can start using your AI application straight away!
/ Building a model quick and easy, with the use of a CNN snippet.
Want to get started with creating your own MNIST classification model? Click here for the full-text version of the tutorial.
- 01/ One day in the life of the U.S. postal service — U.S. Postal Facts