Projects
Here are some of the projects that I worked on in the recent times.
SafetyEye
A computer vision based object detection model tuned to recognise human beings that helps to monitor and detect social distancing in public places. This application uses the YOLO architecture to detect human beings accurately and monitor the distance between people.
Covid Resources website
A web application which has a functionality of displaying tweets regarding covid resources. This application uses flask as the back-end. The scraping of tweets is taken care of a python script written using tweepy.
The website is also hosted on link
Blogator
A web application which has a basic functionality of a blog-aggregator. This application uses RSS feeds to collect content from different blogs or news websites and display new articles on one plate. The news feed will update on a regular basis. There are five categories: general, business, Technology, Sports and Entertainment. I used PHP for using RSS feeds and XAMPP to host a local server and run the application. I developed the front end of the website using HTML and CSS.
ASL Recogniser
A deep learning project which can convert American Sign Language to text. The main intention of this project is to break the communication barrier between an ordinary person and a person with inability to speak or hear. I used transfer learning techniques to train a VGG-19 model and got an accuracy around 99%. Then with the help of OpenCV, I tested this model in real time using a webcam. The prediction done by the model will be displayed on the screen.
Hackathons
Apart from working on personal projects, I also participated in number of Hackathons ans also won a few of them. Here are some of the projects that won prizes in some popular Hackathons
Perfect-pause
Computer Vision based attention monitoring system to aid your movie-watching experience. Link to the project
Here is a youtube video demonstrating our project
Signlingo
Signlingo aims to automate the process of teaching sign language and make it more engaging by gamifying it. The number of sign-language teachers out there is very low compared to the number of learners, moreover they usually charge money for sessions, our solution attemtpts to make the process of learning sign language accessible to everyone for free. Our solution takes input from the webcam, and a custom PyTorch based Convolutional Neural Network determines if the user is showing the correct sign. Here is a youtube video demonstrating our project.