Python, Forensics and Deep learning.A True Story.

ryan dsouza (~ryan50)


Description:

Everyday millions of people submit their identity cards to verify themselves, this may be to log on to the latest payment app or find their better half on a Matrimony Website. For Any business knowing their customer is the foundation for any future transaction with their customer be it in the Advertisement they recommend or the loan they decide not to sanction. But how do you make it happen with the least amount of effort(User Experience) while being reliable(Rejecting Fraudulent Applications) .Python with Deep learning are enabling things which were just not possible a few years ago.

What happens behind the scenes when you provide an identity card to verify yourself?

In this talk we will take a tour of the recent advances made in Computer Vision/Deep Learning through the lens of an Identity Card(Pick Your Favourite Identity Card- Aadhaar, Pan, Drivers Licence etc) .We explore what information you can get from a Single identity card.And while answering this question we will explore ideas like Object detection, Object classification, Face detection, Face recognition.Text recognition and Extraction. And also once we have all this information, How do we know if the card is Genuine.Yes, Forgery detection!

All this and more.

Talk Outline

This talk will be structured as a back and forth between motivating ideas and application of those ideas.

Problem 1

-What is Deep Learning?

  • Overview of recent advances in Deep Learning

Problem 2

-Given an Identity card Classify it

  • What is object classification and why is it important

Problem 3

-If there is a Face in the identity card detect and Recognize it

  • How is Face detection done, and what is face recognition?

Problem 3

-If there is text in the identity card, Extract the relevant fields.

  • What is Language modelling in the context of Text extraction?

Prerequisites:

  1. Fundamentals of python programming
  2. Fundamentals of computer vision

Content URLs:

Slides for the talk - https://docs.google.com/presentation/d/1ZYqTugo3Umd3b7WF15R4wlc7ynuG3KaatowfZTd72So/edit?usp=sharing (These are still work in progress, i am updating them everyday)

Video Description-https://drive.google.com/file/d/1idmq7ssFjWAI6F0by9GZAG5x0Y7SdqYR/view?usp=sharing

These are a few resources on Object classification/Detection.

https://blogs.sas.com/content/subconsciousmusings/2018/11/19/understanding-object-detection-in-deep-learning/

https://tryolabs.com/blog/2017/08/30/object-detection-an-overview-in-the-age-of-deep-learning/

Speaker Info:

After holding research positions in DeepMind, OpenAI and going to found the leading lab to pursue general intell... Ah, Just kidding.

Currently working as a Deep Learning Engineer At Signzy.

Speaker Links:

https://github.com/DecentMakeover

Id: 1273
Section: Data Science, Machine Learning and AI
Type: Talks
Target Audience: Beginner
Last Updated: