Metode Pose Estimation Dalam Sistem Penilaian Seni Tunggal
Currently, the assessment of the seni tunggal movements in pencak silat still uses manual assessment that depends on judges, making it vulnerable to subjectivity and inconsistency. This research aims to build a web-based automated assessment system that utilizes Human Pose Estimation to extract 33 body landmarks from the camera in real time. The landmarks are transformed into 15 joint angle parameters that are sequentially matched with data from the movements in the reference, The final score is calculated from three components: 1. Angle (60%), 2. Transition (25%), and 3. Duration (15%), with adjustments based on the level of completion of the movement sequence (completion). The test results show that 1,845 out of 1,867 reference frames (98.8%) have valid angles, and all angle parameters exceed the validity threshold of 70%. In the score testing, the system distinguishes highly matched movements (83.55), partially matched movements (30.1), and unmatched movements (15), while 13 out of 14 functional test scenarios were successful. The system has the potential to become a tool to assist judges, but validation against human judges' assessments has not yet been conducted.
Authors
- Halim Maulana (ORCID: https://orcid.org/0000-0003-4361-6377)
- Fitra Batubara
Publication Details
- Journal
- Jurnal Teknologi informasi dan Ilmu Komputer
- Published
- 2026-11-22
- DOI
- https://doi.org/10.65258/jutekom.v2.i4.78
- Primary Topic
- Human Pose and Action Recognition
- Type
- article
- Field-Weighted Citation Impact
- 0.00