Sign language learning can support more inclusive communication, yet learning materials based primarily on books or videos may offer limited interaction and immediate feedback. This study presents ASTAKANDA, a gamified web platform developed using Laravel 13 that combines interactive learning activities with real-time hand landmark detection using MediaPipe and K-Nearest Neighbors (KNN) classification based on landmark vectors. The platform also supports the contribution of additional user samples to enrich its dataset. The recognition model was evaluated using 1,201 test samples and achieved 100% accuracy, with precision, recall, and F1 scores ranging from 0.98 to 1.00. Performance testing recorded an average frame rate of 28.2 frames per second (FPS) and GPU memory usage of 1.7 MB. All nine core scenarios in black-box testing passed. These results demonstrate the technical feasibility of integrating real-time sign recognition with gamified learning activities and user-contributed data. However, the findings do not establish the system’s generalizability across users or signing conditions, or its effectiveness in improving learning outcomes. Further testing with independent participants and measures of learning performance is needed.
Keywords
Sign Language, Gamification, Real-Time Feedback, K-Nearest Neighbors, Interactive Learning, Dataset Enrichment.