Detecting Pine Trees Damaged by Wilt Disease Using Deep Learning Techniques Applied to Multi-Spectral Images

Pine wilt disease (PWD) is responsible for significant damage to East Asia’s pine forests, including those in Korea, Japan, and China. Preventing the spread of requires early detection removal damaged trees. This paper proposes a method detecting disease-damaged pines using ortho-images corrected from 5-band multi-spectral images captured by unmanned aviation vehicles. The proposed relies on ResNet18 backbone network connected modified DenseNet module, classifies multispectral (RGB, NIR, Red_Edge) ortho-image patches, visualizes results as heat map. patch-based classifier was retrained with hard negative examples, after which it achieved 98.66% accuracy, an improvement over 96.0% accuracy associated same applied RGB images. resulting map reflects approximate distribution, movement disease. Disease locations are also predicted local maximums When distance between ground truth location less than visible distance, e.g. about 5m, counted correct detection. consists generation followed localization achieves Recall 93.39%, Precision 88.26%, F1-score 90.75%.