Classroom +
About
Classroom + is an AI-based software that can judge anyone's emotions and mood with just facial snapshots. All you need to do is run the application on a supported PC with a camera and software does the magic. It identifies faces and displays their emotion with the level of accuracy. And it does all this with no manual intervention! The software uses deep learning to do the magic and help teachers!
Why build Classroom +
Extended research was conducted to find out ways to see classroom behaviour and attendance without manual interference. The class behaviour survey was dependent on the teacher’s teachings to figure out does it interest the student. Also in many colleges, universities and sometimes school children were bunking classes and teachers weren’t able to figure it out at times, this was misleading teachers at times. So our team decided to take up Artificial Intelligence to improvise teaching and learning for educators and students. We chose artificial intelligence because the chances of it making errors are very few and it has high accuracy. Moreover, in the future artificial intelligence can also predict what will be the child’s behaviour at a particular time and the number of holidays a child will take in a year and when (Only possible after analysing data after about 1 year). It also helps the teachers, parents and school management to figure out how a class is and how much attention a child pays in a particular class. Our prototype is like a high-end surveillance camera. We then researched how to develop it and then we developed the assets gradually.
Why Classroom + is revolutionary
Classroom + is based on facial recognition technology, the same technology that is making the world more secure. The same technology that you probably used to unlock your device. Schools on a daily basis struggle with attendance, and teachers struggle with not being able to read facial expressions. And we are here to just help solve that! This is not it, our software can also be tweaked to become a burglary alarm system! The system would just need the end-user to add images of trusted people and every time our software detects an unknown face with no known face around it triggers an alarm. How simple, yet how secure is that? Our software can also be used in workplaces to detect the emotions of employees. This will not only help the human resources department to assess the happiness quotient of the employees but, to actually work on improving. Possibly boosting employee morale! Our software checkmarks simplicity while making the maximum impact!
Technical Specs
Our project was solely based on Python and we utilised a number of sources libraries to make this project happen. We used some basic libraries like Panadas in addition to Numpy and SciPy to convert these images to matrices and, then we used OpenCV for image processing. The brain of the project was TensorFlow, which was responsible for training the project and analysing real-life images before the product was ready to use. The whole system was based on Python 3.6.7 and TensorFlow version 1.15. In addition to this, we used 3 more libraries. Namely, Haar Cascade, PIL and Keras. The database we used was custom-made and utilised over 600 facial pictures spanned over 5 different moods. The database has been made private due to privacy concerns.