Arvin26/Cattle-detection-model

Beef Cattle Detection Model Weights

This repository hosts trained model weights and selected experiment outputs associated with the research project:

• GitHub project: https://github.com/jianglong26/Beef-Cattle-Growth-Analysis
• Related paper DOI: https://doi.org/10.1016/j.compag.2026.111559

Overview

These weights are used in a beef-cattle monitoring pipeline:

1. Detect cattle instances in aerial farm images (YOLOv11-based detection).
2. Focus on standing cattle instances for downstream processing.
3. Combine detection outputs with segmentation/post-processing in the main project.
4. Extract morphology features and support growth-curve analysis.

Available Weights

• cattledetect_single/: Cattle, single class.
• cattle_detect_double/: Standing and Lying, two calsses.

Folder names and files are kept consistent with the source research project for reproducibility.

Label Semantics

The primary class mapping used in the project is:

• 0: laying
• 1: standing

Please verify class mapping against the exact experiment config before final reporting.

Intended Use

• Research and academic reproduction.
• Benchmarking and method comparison for cattle detection under similar data conditions.

Not intended for:

• Clinical or veterinary diagnosis.
• Safety-critical automated decisions without human validation.

Quick Inference Example

from ultralytics import YOLO

model = YOLO("path/to/best.pt")
results = model.predict(source="path/to/image.jpg", conf=0.25)

For full training/evaluation/inference scripts, use the main project repository.

Citation

If you use these weights, please cite the related paper:

@article{beefcattle2026,
    title   = {Beef cattle growth analysis from aerial imagery with detection and downstream morphology modeling},
    journal = {Computers and Electronics in Agriculture},
    year    = {2026},
    doi     = {10.1016/j.compag.2026.111559}
}

And include the project repository:

• https://github.com/jianglong26/Beef-Cattle-Growth-Analysis

License

This model repository is released under CC-BY-4.0.