The potential of students while they are still in college is largely untapped. Recently, we partnered with The Lakshya Foundation for the NIT Warangal hackathon. We gave students a very interesting problem to solve - face detection - and we are thrilled with their performance.
At Almabase, we want to use the technology to make it as easy as possible for a college to build a thriving alumni fraternity for themselves.
One of the mundane task that we want to automate is information retrieval from old year books through Image Analysis and Optical Character recognition algorithms. A lot of colleges print yearbooks for every graduating batch and these contain some very useful information - old profile pictures, name, address etc. Instead of extracting these information manually, we wanted to build technology to do it.
Picture below is a typical yearbook page
We want to start with extracting profile pictures out of this image. The first step towards that is detecting faces.
Once we presented the problem statement for the hackathon, 8 teams registered for it. None of the students had ever worked on image analysis before. We then had a couple of calls where we gave them some inputs and sent some reading material. That’s all they needed to understand how to go about it. On the day of the hackathon, we guided all the teams and in under 18 hours 4 teams had working solutions - 1 in matlab and 3 in python using machine learning.
This is a sample of the input and the output(Notice the rectangles around the faces.:
Their solutions even worked real time! This is picture of a team trying it out with their webcam.
Most importantly, they all learnt a lot and had fun while doing this.
Overall, a great experience for us and reinforced our belief in the potential of our colleges.