Tall Fescue Abundance Classification in Kentucky Pastures Using Deep Learning

A deep learning model achieved 96% classification accuracy across three abundance classes for tall fescue. The dataset supports research into automating forage management and assessing endophyte-related toxicity risk for livestock. It was created by Ross E. Binkley and last updated in April 2026.

Use Cases

• Automating tall fescue abundance estimation based on red-green-blue pasture images.
• Comparing traditional visual estimation methods with deep learning approaches like YOLO and ResNet.
• Supporting risk assessment for livestock toxicity by detecting the spatial distribution of tall fescue.
• Developing continuous monitoring applications for forage management to reduce manual labor.

Strengths

• Model accuracy of 96% is reported for the YOLOv8x architecture.
• Ground-truth data was evaluated using a traditional occupancy grid method.
• The dataset is licensed under CC-BY-4.0 for open use.

Limitations

• The dataset description does not specify the number of images or rows of data.
• Column-level documentation is absent; field semantics must be inferred after download.
• Geographic coverage is limited to pasture systems in Kentucky.

Provenance

Source

Ross E. Binkley via figshare.

Collection Method

Ground-truth abundance classes were compared against estimates derived from red-green-blue images using convolutional neural networks.

Freshness

Last updated 2026-04-23 05:34:41; freshness should be verified.

Geography

Pasture systems in Kentucky.