Indonesian Sign Language (SIBI) Detection: A Landmark-Based Approach with Scikit-Learn Utilizing Random Forest Classifier

This paper discusses computer vision-based sign language by utilizing landmark detection and Scikit-Learn. The objective of this research is to develop a model that proficient in the accurate recognition of hand gestures specific to sign language. Pictures of hand gestures are taken using a camera and hand landmark are detected by utilizing MediaPipe. Following to landmark detection, a machine learning model is developed using Scikit-Learn employing features extracted from these landmarks. Evaluation of the model using the Indonesian sing language (SIBI) yielded competitive result, proving the efficacy of the proposed model without specific hardware. This work significantly improves the communication technology for the deaf, leveraging computer vision to broaden the research opportunity in sign language recognition utilizing available equipment. The accuracy obtained from this model is $100 \%$ which means the classification is correct, and reinforces the effectiveness of the approach to the recognition and classification of hand movements in sign language.