The problem Shikshak solves

During this pandemic, the online classes and the work from home charades easily consume over 3GB of a person's network data. This might seem normal for a person middle-class or above, but the weekly cost of such a high bandwidth is not viable for everyone, especially for the ones looking for affordable education. This is where "Shikshak" helps the needy. We provide a low-bandwidth solution to attending online classes through our portal. The magic happens in how we transmit the image of the board on which the teacher is writing. We heavily compress it to the format such that there is almost an 85% decrease in internet consumption using our product. From the machine learning perspective following are some challenges we faced: detect corners of the blackboard, make a suitable boundary of the best-suited blackboard as understood by machine learning, define final edges of the blackboard, dot map the pixels to understand the written content on the blackboard.

Challenges we ran into

The major hurdle on the web development side was to configure webRTC in such a way so that students can only use an audio channel for real-time communication with the teacher. The second hurdle was to bring the frames of the teacher's video down to such a format so that net consumption can be decreased.