Cochineal Colony Detection in Cactus Pear: A Deep Learning Approach

Object detection is a fundamental task in computer vision, playing a crucial role in various applications such as surveillance, autonomous driving, and agriculture. In agricultural contexts, object detection techniques are essential for assessing plant health and implementing targeted interventions. In this paper, we introduce a novel methodology for the detection of cochineal colonies of Dactylopius opuntiae in cactus pear, which aims to estimate the degree of infestation and facilitate precise treatment strategies. Leveraging recent advancements in deep learning, we present a new dataset specifically curated for colony cochineal detection in cactus pear. We evaluate the performance of three state-of-the-art deep learning models, namely YOLOV7, YOLOV8, and YOLO-NAS, using our dataset. Through rigorous experimentation and comparative analysis, we identify YOLOV8 as the most effective model for colony cochineal detection in cactus pear. The proposed approach not only offers accurate colony detection but also provides valuable insights for implementing precise treatment measures, thereby contributing to the efficient management of plant infestations.