YOLO v5, v7 and v8: A Performance Comparison for Tobacco Detection in Field

Accurate tobacco plant detection is crucial for effective agricultural management strategies. Due to the challenges posed by high plant variability, adverse lighting conditions, and occlusions, existing machine learning algorithms and convolutional neural networks (CNNs) have struggled to achieve satisfactory detection accuracy under real-world scenarios. This paper employs YOLOv8, a state-of-the-art object detection algorithm, to reliably detect tobacco plants. A comprehensive evaluation and benchmark of 12 state-of-the-art YOLO object detection algorithms including the YOLOv8, is established for tobacco detection. YOLOv8 demonstrates remarkable test accuracy of 96.4compared to YOLOv5 and YOLOv7. Performance comparisons with YOLOv5 and YOLOv7, two ancestor algorithms, using test images further solidify YOLOv8’s superior detection capabilities for tobacco plants. YOLOv8’s real-time object detection ability renders it an ideal solution for mobile and embedded devices, opening new possibilities for on-the-go agricultural management.