Parato, Lei I. creator author. Braga, Angelika V. author. Villamer, Jemima R. author. Oñate, Joseph Jessie S., adviser. Prianes, Freddie B., chair, panelist. Benitez, Ian P., panelist.. Onesa, Rosel O., panelist. text bibliography ph 2024 monographic eng

print

180 leaves. Filipino Sign Language (FSL) is the primary communication method for the Deaf com- munity in the Philippines, however limited knowledge of the language among the hearing population creates significant communication barriers. This study addresses this issue by developing a real-time FSL-to-text translation system. While deep learning has shown po- tential in gesture recognition, the lack of high-quality, large-scale FSL datasets hinders progress in this field. To bridge this gap, the study aimed to create a culturally authen- tic FSL dataset, build an accurate recognition model, and develop a system to deal with the current gaps in FSL research. A dataset of 846 video clips featuring 118 FSL ges- tures was created in collaboration with the Deaf community, ensuring linguistic accuracy. The system employed a Faster R-CNN model for gesture recognition, using a server-client architecture for real-time processing. Each frame of the video were extracted using Con- volutional Neural Networks, while Faster R-CNN and Gated Recurrent Units classified the signs. The resulting system achieved 90% accuracy with strong precision, recall, and F1-scores, demonstrating its potential to address communication challenges for the Deaf community. Lei I. Parato, Angelika V. Braga, and Jemima R. Villamer. Undergraduate Theses (BSCS) - Computer Science. Camarines Sur Polytechnic Colleges, 2024. Includes bibliographical references. CSPC 261021 20261021133020.0 eng