CNN-lytical


  • Web Development


Mentors :

  • Akshat Kumar

  • Vaibhav Raj

  • Ashwin Ramachandran

Mentees :

  • 15


The goal of this project is to make you well-versed with CNNs, enough that you can write an entire network in PyTorch from scratch. It will kickstart your journey into Deep Learning.
Prerequisite: Well-versed with Python and Selection Assignment: https://docs.google.com/document/d/13QSgl5aUxiG2sLNOjez2j3q8GR0gzEoppl17ORRSSWQ/edit?usp=sharing

Tentative Timeline :

Week Work
Week 1-2 Introduction to Neural Nets, code up Multi-Layer Perceptron from scratch in Numpy (in a modular fashion) - Lectures 4, 5, 6
Week 3 Introduction to PyTorch, training NN for MNIST classifier.
Week 4 Convolutional Neural Networks in PyTorch - Watch lecture (CS231n Lecture 7) and learn PyTorch implementation.
Week 5 YOLO (YOLO paper) implementation.
Week 6 UNet for Segmentation - Watch implementation at U-Net from scratch using PyTorch, read the paper at U-Net paper
Week 7-12 (i) GradCAM (ii) Self-Driving Cars Simulation and more....