Research Assistant at Computer Science cs-090326

Project title:

Deep Learning Based Early Detection of Mastitis in Dairy Cows to Improve the Production of Quality of Milk in Oman Using Infrared Thermography and Computer Vision

Project Description:

This project addresses mastitis a devastating inflammatory condition threatening Oman’s dairy industry by employing artificial intelligence and infrared thermography for early, non-invasive detection. The system will collect and analyze thermal images of cows’ udders and eyes under standardized conditions, training deep learning models to detect subtle temperature anomalies indicative of subclinical mastitis before visible symptoms appear. This scalable, technology-driven solution will enable timely intervention, reduce economic losses, improve milk qualityand animal welfare, and support Oman’s Vision 2040 goals for agricultural innovation and food security.

RA Duties and Responsibilities:

Role: Data Acquisition & System Development Lead.

• Phase 1: Experimental Setup & Data Collection (Months 1–3)
  • Coordinate with dairy farms and veterinarians for scheduling and sample access.
  • Assist in installation and calibration of infrared thermography and RGB cameras.
  • Develop standardized image acquisition protocol.
  • Collect thermal and RGB images during milking sessions.
  • Record metadata (cow ID, lactation stage, milk yield, SCC results).
  • Ensure secure storage and backup of collected data.

• Phase 2: Data Processing & Model Development (Months 3–5)
  • Preprocess thermal and RGB images (normalization, segmentation of udder region).
  • Develop structured multimodal dataset.
  • Implement baseline deep learning models (CNN, ResNet, EfficientNet).
  • Develop multimodal fusion architecture (early fusion / late fusion).
  • Optimize hyperparameters and training performance.

• Phase 3: System Integration & Validation (Months 5–7)
  • Develop prototype early detection system.
  • Implement real-time inference pipeline.
  • Conduct validation experiments using unseen data.
  • Optimize computational efficiency for deployment.
  • Contribute to preparation of system architecture diagrams for publication.

• Expected Deliverables by Month 7
  • Multimodal Mastitis Image Dataset (Thermal + RGB + Clinical Labels)
  • Deep Learning-Based Early Detection Prototype System.
  • Performance Evaluation Report.
  • At least One Journal/Conference Paper Submission.
  • Final Technical Report for the Funding Agency.

Qualification/degree:

• Bachelor’s in Computer Science