Cauliflower Disease Detection and Field Mapping System Using Deep Learning

This project introduces a scalable and intelligent system for detecting and mapping diseases in cauliflower crops using deep learning. Instead of relying on real-time inference, which often demands high-end hardware and suffers from latency in field conditions, this system adopts an offline field mapping approach. Images of cauliflower plants are captured systematically across the field, with each image tagged according to its row-column location to represent the spatial layout.

Disease detection is performed using advanced object detection models — YOLOv8, YOLOv9, and YOLOv11 — which have been trained or fine-tuned to recognize key cauliflower diseases: bacterial spot rot, black rot, downy mildew, and no disease. These models process the images offline to ensure high accuracy and avoid the constraints of in-field inference.

The prediction results, along with their associated grid locations, are stored in a SQL database, forming a structured dataset that represents the health status of each plant across the field. This data is then used to generate an interactive HTML-based field map, where each plant's status is visualized using color-coded markers, allowing for easy interpretation of disease distribution and severity.

This system offers several benefits:

·         It supports batch processing of images, allowing flexible scheduling and better accuracy through ensemble model validation.

·         It provides actionable insights to farmers and agronomists, helping in targeted interventions and efficient disease management.

·         It enables historical tracking of disease patterns through SQL-based record storage.

·         In summary, this project provides a cost-effective, field-adapted solution for cauliflower disease detection and visualization, bridging the gap between AI-powered diagnostics and practical agricultural needs.

Keywords: cauliflower disease detection, deep learning, YOLOv8, YOLOv9, YOLOv11, object detection, offline field mapping, SQL database, interactive field map, disease visualization, batch processing, ensemble model validation, targeted interventions, agricultural AI diagnostics.