Integration of Artificial Intelligence and ROS2 for Automated Basil Detection in Smart Agriculture
This project focuses on developing an AI-based perception module for automated basil plant detection. It includes dataset preparation, YOLO model training, and ROS2 integration for live detection, localization, data storage, and visualization within the AgriRobot system.
Project Highlights:
Dataset Development
Collection and annotation of 800 greenhouse images (7,106 basil plants) using Roboflow.
Collection and annotation of 800 greenhouse images (7,106 basil plants) using Roboflow.
YOLO Model Comparison
Training and benchmarking of five lightweight YOLO variants for accuracy and speed.
Training and benchmarking of five lightweight YOLO variants for accuracy and speed.
Embedded Optimization
ONNX and NCNN optimization, reaching real-time detection on Raspberry Pi 5.
ONNX and NCNN optimization, reaching real-time detection on Raspberry Pi 5.
ROS2 Live Integration
Detection, localization, CSV storage, and 2D/3D visualization nodes.
Detection, localization, CSV storage, and 2D/3D visualization nodes.
