WeedScout: Real-Time Autonomous blackgrass Classification and Mapping using dedicated hardware

TL;DR

WeedScout addresses real-time detection and mapping of the problematic weed Alopecurus myosuroides (blackgrass) in wheat fields by deploying edge-accelerated YOLO-based detectors on a NVIDIA Jetson Nano platform to infer blackgrass density and provide density overlays offline. The methodology combines two YOLO variants (YOLOv8 and YOLO-NAS) trained on two curated datasets, with data augmentation and an end-to-end pipeline that operates on embedded hardware without requiring continuous internet connectivity. Key contributions include the release of white-box datasets and model weights, an edge-optimized workflow, and a comparative analysis showing YOLOv8 generally offers better performance and inference speed on the NJN, albeit with challenges in low/medium density detection due to dataset bias. The work demonstrates the practical viability and current limitations of deploying real-time precision weed management on inexpensive edge hardware, and outlines directions (data diversification, advanced runtimes like TensorRT/DeepStream) to reach real-time fps and enable targeted herbicide or mechanical weed control.

Abstract

Blackgrass (Alopecurus myosuroides) is a competitive weed that has wide-ranging impacts on food security by reducing crop yields and increasing cultivation costs. In addition to the financial burden on agriculture, the application of herbicides as a preventive to blackgrass can negatively affect access to clean water and sanitation. The WeedScout project introduces a Real-Rime Autonomous Black-Grass Classification and Mapping (RT-ABGCM), a cutting-edge solution tailored for real-time detection of blackgrass, for precision weed management practices. Leveraging Artificial Intelligence (AI) algorithms, the system processes live image feeds, infers blackgrass density, and covers two stages of maturation. The research investigates the deployment of You Only Look Once (YOLO) models, specifically the streamlined YOLOv8 and YOLO-NAS, accelerated at the edge with the NVIDIA Jetson Nano (NJN). By optimising inference speed and model performance, the project advances the integration of AI into agricultural practices, offering potential solutions to challenges such as herbicide resistance and environmental impact. Additionally, two datasets and model weights are made available to the research community, facilitating further advancements in weed detection and precision farming technologies.

Figures (4)

• Figure 1: NVIDIA Jetson Nano.
• Figure 2:
• Figure 4: The proposed WeedScout architecture workflow.
• Figure 5: Blackgrass detection overlay initial design. Yellow is being used to demonstrate low density and orange is being used to demonstrate medium density. High density patches would have a red overlay.