I successfully completed Term 1 of Udacity’s SDCND! What a grind for three months – a thrilling grind though. It was extremely satisfying to receive the final email saying I passed the last project of the term. The past two weeks, I’ve spent relaxing and taking some time away from programming. Well, somewhat of a break. I started Udacity’s C++ primer to get ready for Terms 2 and 3, which primarily focus on utilizing C++.
During the end of September and beginning of October, I turned in two separate projects (here and here). I was a bit surprised to see us stray away from neural networks so quickly during the end of Term 1, but it was valuable none the less. We utilized computer vision algorithms and some rudimentary machine learning to detect lane lines and cars.
I would have loved to approach both of these projects using Deep Learning, but stuck with Udacity’s recommendations this time around. In the future, if I have time, I’ll revisit them to attempt using Deep Learning instead. One particularly interesting approach, uses the YOLO: Real-Time Object Detection system to detect vehicles. I find this to be the most adequate implementation currently and it was even recommended by my Udacity reviewer. The full paper can be found here. Check out xslittlegrass’s repo and Subodh Malgonde’s repo for examples of YOLO in the Udacity course.
Overall, it’s been an awesome past three months and even year. This time last year, I had just started learning Ruby and Ruby on Rails. I was barely scratching the surface of what it meant to write code / build programs. Now, I’ve completed my first term of Udacity’s advanced SDCND and am coding part time at work. Recently, I build a script to parse Active Directory data in Python to help categorize user access rights. Check it out here. I’m also moving along with some work in Alfresco, not the most enlightening work, but it’ll help sharpen up my web dev skills.
Until next time 🙂
P.S. Term 2 starts on November 2nd. Woohoo!