Drone

Monitoring Application Design: Create a framework for agricultural monitoring, focusing on olive tree health, using drones, multispectral cameras, and AI.

Plant

and Disease Recognition AI Model: Develop AI and computer vision models for detecting and classifying tree diseases, especially Xylella, based on multispectral imagery.

Optimizing

Data Acquisition Methodology: Determine the best drone flight conditions and distances for capturing accurate data, improving the AI system’s precision.

Enhanced

Crop Management: Provide actionable insights for farmers to optimize crop management, treatments, and strategies for healthier olive trees.

SENSOR 2.0 by TAAL

To develop an advanced recognition system for olive trees and Xylella detection, enhancing agricultural management and improving olive crop health and yield.

waves
  • DJI Mavic 3M
  • High-Resolution RGB Cameras
  • Multispectral Sensors
  • DJI Terra: Software for 3D model reconstruction and 2D NDVI map generation, supporting the training of object detection algorithms.
  • CloudCompare: 3D point cloud processing for background removal and object labeling using connected component analysis.

Outcomes

Definition of the most appropriate drone altitude for multispectral data acquisition

Trees detection model on data acquired by drone’s camera

Health status classification model of olive trees by using RGB images and NDVI image index

Challenges Faced

  1. NDVI Information: Higher altitudes reduce the amount of useful information for tree classification due to lower image resolution.
  2. Initial Dataset Issue: The first dataset was collected from an area with all dead trees, making it unsuitable for training detection and classification models.
  3. 3D Model Generation: Creating 3D models from drone images was time and resource-intensive. Background removal for training data relied on DJI Terra and Cloud Compare software for processing.
  4. Preprocessing Challenges: Data preprocessing required camera calibration and distortion correction according to DJI specifications, using metadata from the images.
  5. Health Status Model: The model needed a significant amount of supervised data, requiring manual labeling to train accurately.

Impact (Socio-economic & Environmental)

Socio-Economic Impact:

  • Improved Agricultural Productivity: Enhanced olive grove management
  • Job Creation & Skill Development: New tech-oriented job opportunities and skill development
  •  Advanced Agricultural Practices: Introducing cutting-edge drone and AI technologies modernizes farming.