Cotton Growth Stages Detection Using Fine-Tuned YOLOv8 Deep Learning Model

This study focused on detecting different growth stages of cotton include cotton bud, cotton blossom, early cotton boll, split cotton bolls and matured cotton from images captured in the actual field conditions despite the challenges posed by the unstructured and predominantly green environment of the field. Using a custom trained YOLOv8 deep learning model for monitoring cotton and determining their growth stages hold significant value in optimizing agricultural practices. YOLOv8, known for its enhancements in accuracy and speed, serves as a strong foundation for this application. The selection of YOLOv8l as superior model variant for identifying the growth stage of cotton and attaining average mean precision (mAP@0.5) scores of all classes 0.643 with precision, recall, and F1 of 1, 0.8 and 0.64 at confidence 0.942, 0.000, and 0.270 respectively, underscores the effectiveness of these models in accurately identifying cotton plants and assessing their growth stages. The model capability could provide valuable insights for growers, enabling them to make informed decisions regarding plant care, resource allocation, and harvest timing.