DStruct: Handwritten Data Structure Problem Solving Using Detectron2 and YOLO

This project presents a comprehensive approach to address the challenges of handwritten data structure and algorithm (DSA) problems through the integration of computer vision, Optical Character Recognition (OCR), and algorithmic analysis. Our methodology entails the creation of meticulously annotated datasets of handwritten graph images, crucial for training our models effectively. Leveraging state-of-the-art techniques, we employ Detectron2 and YOLO for robust detection of handwritten elements, including nodes and edges in DSA problem graphs. Additionally, the Microsoft Azure OCR API facilitates the accurate extraction of textual content from handwritten images, providing essential data for algorithmic analysis. Through experimentation and evaluation, our approach demonstrates remarkable accuracy and efficiency gains compared to conventional methods, with accuracy rates about 95% in adjacency matrix generation and 81% in weighted matrix generation. These results underscore the efficacy of our methodology in streamlining algorithmic problem-solving processes, significantly reducing manual effort, and enhancing productivity. The implications of our findings extend beyond mere efficiency gains, as the automation of DSA problem analysis holds promise for accelerating learning and education in the field. Future work will focus on refining our models, expanding to tackle diverse DSA problems, and exploring educational applications to empower learners and educators alike in the realm of data structures and algorithms.