USU-Corn-WeedDB: A UAV RGB Image Dataset for Multi-Species Weed Detection in Forage Corn

2026-06-08 · Source: cs.CV updates on arXiv.org · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems · Depth: Expert, short

Summary

USU-Corn-WeedDB is a new publicly available UAV RGB image dataset designed for multi-species weed detection in forage corn, addressing the scarcity of field-representative training data for site-specific weed management. Collected from a commercial forage corn field in Cache Valley, Utah, on June 27, 2025, using an Autel EVO II Dual 640T V2 drone at approximately 10m above ground level, it offers a ground sampling distance of 0.48 cm/pixel. The dataset comprises 366 full-resolution images, tiled into 8,800 patches at 640x640-pixel resolution. Of these, 800 images contain 10,539 bounding-box instances manually annotated for common lambsquarters, redroot pigweed, and green foxtail, with redroot pigweed constituting 53.86% to reflect natural class imbalance. The remaining 8,000 tiles are an unlabeled pool for semi-supervised learning. Validation with 28 object detection models, including YOLOv8 through YOLO26 and RT-DETR, yielded test set mAP@0.5 scores from 0.773 to 0.840, demonstrating competitive performance from lightweight architectures suitable for edge deployment.

Key takeaway

For Computer Vision Engineers developing agricultural robotics or site-specific weed management systems, USU-Corn-WeedDB offers a critical, field-representative dataset to overcome data scarcity. You should leverage this public resource to train and validate multi-species weed detection models, particularly exploring semi-supervised learning with its extensive unlabeled image pool. Prioritize lightweight architectures like YOLO variants, as they demonstrated competitive mAP@0.5 scores (0.773-0.840) suitable for efficient edge deployment on UAVs.

Key insights

A new UAV RGB image dataset addresses data scarcity for multi-species weed detection in forage corn.

Principles

• Preserve natural class imbalance in datasets to reflect real conditions
• Lightweight models offer competitive performance for edge-deployed UAV systems

Method

Acquire UAV RGB imagery, tile full-resolution images into patches, manually annotate specific weed species with bounding boxes, and reserve unlabeled data for semi-supervised learning.

In practice

• Utilize USU-Corn-WeedDB for training deep learning models in agricultural robotics
• Explore semi-supervised learning with the dataset's unlabeled image pool

Topics

• USU-Corn-WeedDB
• UAV Imagery
• Weed Detection
• Forage Corn
• Object Detection
• Semi-supervised Learning

Best for: AI Scientist, Research Scientist, Computer Vision Engineer

Related on AIssential

• USU-Corn-WeedDB: A UAV RGB Image Dataset for Multi-Species Weed Detection in Forage Corn — A new UAV RGB dataset enables multi-species weed detection in forage corn, supporting supervised and semi-supervised deep learning.
• Carbon Robotics built an AI model that detects and identifies plants — A new AI model enables instant, real-time weed recognition for autonomous agricultural robots, eliminating retraining.
• SWNet: A Cross-Spectral Network for Camouflaged Weed Detection — Cross-spectral fusion and edge refinement are crucial for accurate camouflaged weed detection in agriculture.
• Small Model, Big Results: How a 5MB On-Device NSFW Detector Outperforms Cloud APIs and Passes a… — Small, on-device object detection models can surpass large cloud APIs in NSFW accuracy and fairness by focusing on anatomical shapes.
• Tomato Multi-Angle Multi-Pose Dataset for Fine-Grained Phenotyping — TomatoMAP dataset enables accurate, reproducible fine-grained tomato phenotyping using multi-angle imagery and AI models.
• Comparative Analysis of Military Detection Using Drone Imagery Across Multiple Visual Spectrums — Diverse visual spectrum datasets significantly enhance drone-based military object detection robustness in varied environments.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cs.CV updates on arXiv.org.