SAAIL Lab Publications
Biosystems Engineering · 2019 6.5 194+ Citations Prior Institutional Research · Pre-IITMZ (2019)

A machine vision system for early detection and prediction of sick birds: A broiler chicken model

Cedric Okinda 1 , Mingzhou Lu 1 , Longshen Liu 1 , Innocent Nyalala 1 , Caroline Muneri 1 , Jintao Wang 1 , Hailin Zhang 1 , Mingxia Shen 1
1College of Engineering, Nanjing Agricultural University, Nanjing, China
Biosystems Engineering (Elsevier), Vol. 188, pp. 229–242 (Dec 2019)

Summary: Continuous overhead machine vision posture and mobility tracking for early detection and prediction of illness in commercial broiler houses (194+ citations).

Abstract

Early illness detection prevents devastating poultry flock mortality. Using continuous overhead RGB-depth imaging, shape posture descriptors, and optical-flow mobility features, we construct an automated classifier and early warning system for sick broiler chickens validated against clinical diagnosis.

Keywords: broiler healthmachine visionearly disease detectionBiosystems Engineeringlivestock

Key Findings & Contributions

97.8%
Classification Acc
RBF-SVM (all posture + speed)
97.5%
Shape-Only RBF-SVM
5 posture descriptors (p < 0.05)
Day 4
Earliest Warning
Post-challenge detection
194+ Cites
Biosystems Eng.
Vol. 188, pp. 229-242
  1. Monitored isolated control vs. Newcastle-virus-challenged broiler flocks continuously using overhead RGB-depth imaging and walk-speed extraction.
  2. Established statistical significance (p < 0.05) across five posture descriptors and walk speed, achieving earliest disease detection on Day 4 via circle variance and elongation (Day 6 via eccentricity and walk speed).
  3. Achieved 97.5% accuracy (posture shape only) and 97.8% accuracy (combined shape + mobility) with RBF-SVM for non-intrusive early flock warning.

Benchmark Results

Early Infection Detection Timing & RBF-SVM Classification Accuracy (2019)

Feature GroupDiagnostic Capability / TimingRBF-SVM Accuracy (%)
Circle Variance & ElongationEarliest significant shift at Day 4 post-challengePart of 97.5% posture model
Eccentricity & Walk SpeedSignificant behavioral shift by Day 697.8% (all features combined)
Convexity & ComplexityNo statistically reliable early warningExcluded as standalone signal

BibTeX

@article{okinda2019sickbirds,
  title={A machine vision system for early detection and prediction of sick birds: A broiler chicken model},
  author={Okinda, Cedric and Lu, Mingzhou and Liu, Longshen and Nyalala, Innocent and Muneri, Caroline and Wang, Jintao and Zhang, Hailin and Shen, Mingxia},
  journal={Biosystems Engineering},
  volume={188},
  pages={229--242},
  year={2019},
  doi={10.1016/j.biosystemseng.2019.09.015},
  publisher={Elsevier}
}
APA Citation

Okinda, C., Lu, M., Liu, L., Nyalala, I., Muneri, C., Wang, J., Zhang, H., & Shen, M. (2019). A machine vision system for early detection and prediction of sick birds: A broiler chicken model. Biosystems Engineering, 188, 229–242. https://doi.org/10.1016/j.biosystemseng.2019.09.015