Wood Type Classification Through Image Processing Using YoloV8

Image processing is increasingly prominent in current research and continually evolves with technology. More exotic wood types in Asian countries, such as Kamagong and Sampalok, have received less attention in research papers. This study aims to detect four different types of wood—namely, Kamagong, Agoho, Acacia, and Sampalok—using Yolov8 and Raspberry Pi 4B. The research is significant for exploring Y olov8 in terms of sample size, training data splits, and other factors. For this study, 100 samples per wood type were collected, totaling 400 samples overall. After detection, the study seeks to classify each wood sample into one of the four types. The training images were consistently lit. A multiclass confusion matrix is employed to assess the model's accuracy. Following testing, the device achieved a detection accuracy of 72.50%. Misclassifications primarily stemmed from variations in lighting conditions during result extraction.