IDENTIFIKASI GERAKAN TANGAN PADA SANDI SEMAPHORE PRAMUKA SECARA REALTIME MENGGUNAKAN DECISION TREE

Identifying hand gestures in semaphore code accurately and in real time is a challenge. Especially for Scouts who are just learning this skill to minimize errors that can result in inappropriate information received and can affect the safety and effectiveness of communication. The use of Decision Tree in identifying hand gestures can make a significant contribution for Scouts to communicate more effectively. Based on the test results, this model can recognize letter classes in semaphore ciphers with normal lighting as evidenced by a higher accuracy rate. The average accuracy in normal light is 94%. In low-light conditions, it showed lower performance. In the first test, the model achieved 74% accuracy by recognizing 20 classes, while in the second test, the accuracy dropped to 66% by recognizing 18 classes. Confusion matrix testing is used to evaluate the Accuracy, Recall, and Precision levels in model training using Decision Tree.