Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{202773,
        author = {Dr. Nazirkar S. B. and Gaikwad Tejaswini Manik and Sakshi Appaso Kumbhar and Sakshi Gorakh Babar and Dr. Shah Saloni Niranjan},
        title = {Classification of Sign Language Characters by Applying a Deep Convolutional Neural Network},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {12},
        number = {12},
        pages = {8956-8960},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=202773},
        abstract = {Communication barriers between the deaf and hard-of-hearing community and the general population remain a significant challenge in everyday interactions. Traditional methods of sign language interpretation often require human interpreters, which may not always be available, leading to delays and limited accessibility. To address this issue, the Sign Language Detection System proposes an intelligent, real-time solution that utilizes computer vision and artificial intelligence to translate sign language gestures into text and speech. The system captures live video input through a webcam and processes it using advanced hand tracking techniques powered by MediaPipe to extract precise hand landmarks. These landmarks are then analyzed using a dual mode classification approach that combines a deep learning model with a rule-based heuristic engine. This hybrid mechanism ensures high reliability and continuous operation, even in cases where the trained model is unavailable or produces low-confidence predictions. To enhance accuracy and stability, the system incorporates prediction smoothing techniques and a gesture hold-time mechanism, reducing false detections caused by rapid or transitional hand movements. The recognized gestures are converted into readable text and simultaneously transformed into audible speech using a multi-threaded text-to-speech engine, enabling seamless real-time feedback without interrupting system performance. The application is deployed using a Flask-based backend and is containerized with Docker to ensure scalability, portability, and ease of deployment across different platforms. Additionally, the system supports a flexible training pipeline that allows users to collect custom gesture datasets and improve model performance over time. The proposed system provides an efficient, scalable, and user-friendly solution for real-time sign language interpretation. It aims to bridge communication gaps, promote inclusivity, and enhance accessibility by offering a reliable and automated platform for gesture-based interaction.},
        keywords = {},
        month = {May},
        }