Project Overview

In this project, we attempt to build a multi-digit classifier using Convolutional Neural Networks and OpenCV for image data.

We initially train our Deep Learning model on the MNIST dataset. We do this as usual by splitting the data into test and train splits.

Image thresholding comparison


At this point we also define the kernels with which we will carry out our morphological transformations. We use both eroding and dilation so as to get a clearer picture of our image.

Multi-digit detection


Finally, we are able to detect multiple digits in images and we are able to detect how many digits the number has.

Head on over to the link below to check out the entire python notebook which contains the entire step by step process of first training the model, applying image transformations on our test image and finally making a prediction.

• Repository