Segmentation of Drone-captured Images for Illegal Cannabis Plantation Detection

Abstract

Throughout 2021, to carry out a government project of creating a thematic chart of illegal cannabis plantation fields, an aerial mapping operation was carried out utilizing drones. To increase the efficiency of detecting an illegal cannabis plantation, a simple and robust image segmentation was used. Thousands of aerial images were segmented and then illegal cannabis plantation fields were detected using the thresholding method. The result was then compared to the illegal cannabis plantation field detected by experts from the Indonesian National Narcotics Agency. Based on this comparison, the algorithm proposed in this work was able to detect Illegal cannabis plantation fields with an accuracy of 65%.