Projects
Group Highlights
2024
These days it is very hard to keep up with all the hackathons happening here out Hackathon Scraper comes to help. We have used tools like selenium to automate a web browser and search hackathon websites and other sources to get information of new Hackathons. Information gathered from these sources is then passed to an LLM model (currently we are using gemini API) to extract data like dates of hackathon and registration link etc.. This information is presented to the user in a structered format through a website.
This project focuses on streamlining the curation of Online Assessment (OA) questions by utilizing Machine Learning techniques to extract text from collected question images and compile them into a structured database. A dedicated web platform has been developed to allow users to upload OA images, which are processed to update and expand our question repository. This initiative aims to provide future students reliable resource for accessing and analyzing previously asked questions in an efficient manner.
2021
This is an implementation of Communication in MARL using Graph Neural Network. This has been trained and tested on StarCraft II, showing improved training and performance metrics throughout all the maps. Implemented on top of PyMARL for easier comparative study with respect to other algorithms or implementations like ePyMARL.
We focus on hate speech detection in multi-modal memes, solving the Facebook Meme Challenge to predict whether a meme is hateful or not. Visual modality is explored using object detection and image captioning to fetch the actual caption.