Deteksi Pengenalan Citra Rumah Adat Timor Leste Menggunakan Yolov6

Abstract

Timor Leste is a country that has a variety of cultures. Apart from local culture, the people of Timor Leste follow local cultural values inherited from their ancestors, one of which is the traditional house. The people of Timor Leste also have several tribes, including the Marobo tribe and the Samoro tribe. Each of these two tribes has its own traditions and customs, including traditional houses. Due to the diversity of tribes in Timor Leste, there are different traditional houses, and each region has its own characteristics. However, in this era of very rapid globalization, Timor Leste's traditional houses are starting to become less widely known by the public. Due to lack of knowledge about Timor Leste traditional houses. One way to introduce Timor Leste traditional houses is by using digital image processing technology. So the author proposes to detect the recognition of Timor Leste traditional houses using the yolov6 method, which is a system that can complete computer vision tasks in real-time object recognition and has fast speed. The best research results with 1500 iterations of training data used were 10 iterations with a fairly high accuracy level of 98.6% with an mAP value of 0.59678. However, it is even better if you use 50 iterations because it has a greater mAP value, namely 0.95612 and a confidence value of 100%.