COT-AD

This paper presents COT-AD, a comprehensive Dataset de-signed to enhance cotton crop analysis through computer vi-sion. Comprising over 25,000 images captured throughout thecotton growth cycle, with 5,000 annotated images, COT-ADincludes aerial imagery for field-scale detection and segmen-tation and high-resolution DSLR images documenting keydiseases. The annotations cover pest and disease recognition,vegetation, and weed analysis, addressing a critical gap incotton-specific agricultural datasets. COT-AD supports taskssuch as classification, segmentation, image restoration, en-hancement, deep generative model-based cotton crop synthesisand early disease management, advancing data-driven cropmanagement.