Deep Learning for Computer Vision, with Soham Ghormade

Soham Ghormade presented at Code & Supply in Pittsburgh, PA. Learn more at http://www.codeandsupply.co

"""
Andrew Ng recently said that AI is the new electricity.

Electricity is transmitted by electric charge. I argue that deep learning is the new electric charge for this new electricity. That is a poor analogy.

But the point is, deep learning is being applied in a diverse set of fields.
In particular, it is used for cancer cell detection, generating art using Neural Style Transfer and for object detection in autonomous vehicles.

In this meetup, we discuss the focus on the object detection and classification part. In particular, we discuss an approach to a digit detection and recognition system. This system takes in a single image and returns any sequence of digits visible in that image.
The sequences of numbers may have varying scales, orientations, and fonts, and maybe arbitrarily positioned in a noisy image. The SVHN dataset (http://ufldl.stanford.edu/housenumbers/) is used as training data.

The goal, from my end, is two-fold:

1. Share the takeaways from my class project on this topic
2. Quickly go over the relevant concepts before sharing the approach for that project

Depending on the interest and if time permits, we can discuss the state of deep learning.
"""

Code examples will be in Python.