Learning 101

Getting Started

OpenCV

Download and set up OpenCV on Linux. Preferably version 4.2 for compatibility with ROS Noetic.

Maths Basics

• What are convolutions? How can it be used to segregate or mix different signals?
• Write a snippet to convolve an $N \times M$ matrix with a $K \times L$ kernel.

Image Processing 101

• Learn about basic thresholding operations, resizing, image erosion, dilation, filters, masking, gaussian thresholding, and image transformations.
• What are the features in an image? Read about and experiment with various feature detection methods.

Task 1

• Design filters to classify the following symbols -

• Write a code to detect these symbols in the image given below -

Object Detection 101

Learn about hue, saturation, contrasts, brightness, and contours, and write a snippet to determine RGB boundaries for an object present in the region of interest.

• Use classic detection techniques to detect a colored object in the live video feed.
• Learn about harr cascading method and implement human face detection.

Into Neural Networks

Write a code that takes numbers from $0$ to $9$ and arithmetic symbols like $-, +, \times, \div$ as input and perform basic calculations.
Input Format: An image of the expression which is to be evaluated.

References

• CS231n Stanford
• Getting started with keras - go through code examples to get a better idea.
• kaggle - refer notebooks and courses for practice
• OpenCV Docs
• Andrew NG
• ERA-IITK Resources