UAV and Camera-Based Weed Detection Performance Metrics Across Three Dates

A 2026 study by Madhusudhan Adhikari compared three UAV-based weed detection methods against a ground-based camera system. The dataset likely contains performance metrics such as prediction accuracy (MAE, RMSE, R²), omission rates, and Bland-Altman analysis results from three mapping dates in a summer fallow field. The data supports research into the operational feasibility of UAV-derived weed maps for site-specific management.

Use Cases

• Compare the accuracy of RGB, NIR, and YOLO-based weed detection methods based on the described performance metrics.
• Analyze the relationship between weed density and prediction bias across different sensing technologies.
• Evaluate the temporal stability of UAV-based weed mapping from early to late season based on the three mapping dates.
• Assess the impact of soil and residue conditions on NIR-based weed detection performance as mentioned in the description.

Strengths

• Includes specific performance metrics: MAE, RMSE, R², omission rates (41–85%), and Bland-Altman bias/LoA.
• Compares three distinct UAV-based methods (RGB ExG, NIR thresholding, YOLOv11) against a camera reference system.
• Data collection spanned three specific mapping dates (including April 24 and June 24) in a summer fallow field.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• The dataset is a 2.1 MB DOCX file, suggesting the primary data may be embedded in a document format rather than a structured data file.

Provenance

Source

figshare, author Madhusudhan Adhikari

Collection Method

UAV imagery collected using RGB and multispectral sensors, with YOLO trained on annotated RGB image tiles, compared to a ground-based camera system.

Time Range

Data collected across three mapping dates, including April 24 and June 24.

Freshness

Last updated 2026-04-30 05:41:34; freshness should be verified.

Geography

Data collected from a summer fallow field; specific location is unknown.