The workshop builds an intuition behind how a digital image is captured, stored and processed. It aims to show what are the traditional and simple object detection mechanisms in Computer Vision and their limitations by examples. Then we show how Machine Learning came to the aid and solved the problems which the traditional CV techniques could not solve.
We will spend time on analyzing the limitations of Machine Learning and how we can address some of these using the Deep Learning techniques. We will dive into the Black box (DL) and try to understand what each layer is doing and so that we can solve problems in an effective manner. We will finally talk about best practises in solving Computer Vision problems, which technique to use, which parameter to tweak, etc.,
The workshop is going to have 3 major parts each with a example problems that we will experiment on, using Jupyter notebooks. At the end of the workshop, each participant should be able to build a network using Keras (Python library for Deep Learning), train and test the model. It is going to be a hands-on and with some mathematics, especially suitable for the beginners to Computer Vision or practitioners who have not had a chance to build from basics.
- Motivation: Interesting applications of Computer Vision
- What is Computer Vision, Machine Vision and Image Processing ?
- Simple Computer Vision based classification (hands-on)
- Machine Learning in CV
- Classification using ML (hands-on)
- Emergence and Dominance of Deep Learning
- Applications of DL (hands-on)
- Compare ML and DL (hands-on)
- How to solve a CV problem by choosing the appropriate technique?