A Sophomore's Guide to the Summer

I can't stress enough how important it is to work on self projects when you're in college. Everything you learn and do then goes a long way. You have all the time in the world, and have the chance to explore as many things as you'd like.

Here's a list of simple projects that I did / wish I did in my 4 years at college.

[OS, OpenGL]
Use opengl to make a city of your file system. Basically, something like nautilus (Ubuntu's file manager GUI) but a 3D model where buildings represent folders and trees represent files. You'll learn a lot of opengl in this which is a really handy tool in the graphics stack, and you'll understand basic file operations. View an implementation by Nikhil Marathe here.

This was something we had to do for the OS course in colg but was a weekend project and it was so much fun. A FAT table visualizer - ie, a visual guide as to how the FAT table changes over course of thousand iterations of rename, creation and deletion of files. View my implementation on Github. I wish I spent more time on this then to make a pretty GUI with fancy looking tables.

[Android, Face Detection]
Another simple weekend project. Use Android's face detection APIs to create a simple face detection app and figure out how many people are on the screen. A really cool app that used this and created something fun is this app on the playstore called Boo. Check it out - its so simple yet so much fun and it'd be a lot of fun to make.

[WhatsApp Usage Logger]
WhatsApp usage logger. There was a point when I was using WhatsApp for almost an hour a day. What I wished to create then was an app that logs how much you've used whatsapp and blocks me from using it after 15 minutes in a day, unless I solve a maths sum, meaning I really want to use it. Again not more than a week's job but you'd understand a lot about Android APIs that are closer to the OS - how app processes are managed, and how to create a service that blocks apps from being opened. I'd love if someone made this and published it. View this and this.

[Computer Vision]
A little deeper in computer vision. Use HoG and SVM to detect any object in a given scene. HoG is like a shape detector. Usually this takes 100ms on our laptop CPUs. That's just 10fps. So, instead of detection in every frame, do it in every xth frame. Use a tracking algorithm like KLT and track in the intermediate frames. This is something I spent 1.5 months on and was really fun.

If you do spend time on any of these or need my help, write to me!
bhardwaj dot rish at gmail dot com.


Sidenote. I really don't understand how spiders/bots can't scrape email addresses if they're written in the form I did above. Silly Internet tradition.


Popular posts from this blog

[Breaking News] AI takes over universe to calculate Pi

Firebase Auth | The Debug vs Release Signature Problem

Finding My Parter