Weed Detection Model Weights (YOLOv7 Variants)

best_yolov7.pt
This file contains the trained weights of the YOLOv7 base model for weed and cotton detection using UAV imagery. The model was trained on annotated agricultural field images to classify and localize cotton plants and weeds with high precision. This checkpoint represents the best-performing epoch based on mAP@0.5 evaluation and achieved the highest weed detection accuracy among the tested variants.

best_yolov7w6.pt
This is the trained weight file of the YOLOv7-w6, a larger and deeper model architecture optimized for higher capacity and better generalization in complex agricultural environments. The model was trained on the same UAV dataset and delivered competitive performance, though it performed weaker than the base model in cluttered and shaded regions.

best_yolov7x.pt
This .pt file contains the trained weights for the YOLOv7-x, an advanced version evaluated our study. It was trained on UAV-captured agricultural imagery for cotton and weed detection and showed balanced detection performance across classes, though with slightly lower weed accuracy compared to the base model. This checkpoint is suitable for tasks requiring robustness and overall detection stability.