Lightweight Multispectral Crop-Weed Segmentation for Precision Agriculture

TL;DR

This paper introduces a lightweight transformer-CNN hybrid model that effectively segments crops and weeds using multispectral imagery, significantly improving accuracy and efficiency for precision agriculture applications.

Contribution

It presents a novel hybrid architecture that processes multispectral data with specialized encoders, achieving high accuracy with fewer parameters for real-time weed management.

Findings

Achieved 78.88% mean IoU on WeedsGalore dataset.

Outperformed RGB-only models by 15.8 percentage points.

Model has only 8.7 million parameters, enabling real-time deployment.

Abstract

Efficient crop-weed segmentation is critical for site-specific weed control in precision agriculture. Conventional CNN-based methods struggle to generalize and rely on RGB imagery, limiting performance under complex field conditions. To address these challenges, we propose a lightweight transformer-CNN hybrid. It processes RGB, Near-Infrared (NIR), and Red-Edge (RE) bands using specialized encoders and dynamic modality integration. Evaluated on the WeedsGalore dataset, the model achieves a segmentation accuracy (mean IoU) of 78.88%, outperforming RGB-only models by 15.8 percentage points. With only 8.7 million parameters, the model offers high accuracy, computational efficiency, and potential for real-time deployment on Unmanned Aerial Vehicles (UAVs) and edge devices, advancing precision weed management.