Fine-Tuning and Accuracy Testing of AI-Driven Livestock Facial Recognition

The objective of this project is to fine-tune and validate the accuracy of FaceIT , an AI-powered livestock facial recognition system. The technology is already developed with established algorithms, but this project will focus on optimizing its performance, improving recognition accuracy, and testing real-world application using a provided dataset of livestock images. Students will work with existing AI models to train, test, and evaluate the system’s effectiveness, exploring possible refinements to enhance its capabilities. Teams are welcome to suggest new approaches for improvement. All intellectual property (IP) remains with PrüvIT Technologies Inc., and participants must sign a Non-Disclosure Agreement (NDA). Tasks and Activities: Dataset Preparation & Preprocessing: Work with provided livestock image datasets, ensuring proper image organization, cleaning, and normalization for AI training. Apply data augmentation techniques (cropping, rotation, contrast adjustment) to improve model robustness. AI Model Training & Fine-Tuning: Optimize hyperparameters, feature extraction methods, and model architectures to improve facial recognition accuracy. Experiment with alternative training techniques, augmentation strategies, or deep learning approaches to enhance detection and identification rates. Model Evaluation & Accuracy Testing: Design structured test cases to assess recognition performance, false positive/negative rates, and model reliability under real-world conditions. Implement a benchmarking framework to compare different training methodologies and quantify model improvements . Reporting & Documentation: Deliver a technical report summarizing refinements, testing methodologies, and results. Provide recommendations for future optimization , including additional data needs or AI architecture improvements. Document all modifications to the FaceIT model and their impact on accuracy.

Development of a Web Application with AI Integration

The objective of this project is to develop a fully functional web application for FaceIT by transforming its Figma-based front-end design into an interactive and responsive user interface. The project will also involve integrating the existing FaceIT backend , ensuring seamless data flow between the front-end and back-end systems. Students will gain hands-on experience in web development, API integration, and UI/UX implementation , contributing to an innovative AI-driven livestock identification technology. Tasks and Activities: Front-End Development: Convert the Figma design into a fully responsive and user-friendly web application using modern web technologies (React, Vue, or similar) . Implement dynamic UI components, dashboards, and data visualization for intuitive user interaction. Backend Integration: Connect the FaceIT backend (already functional) via APIs to enable real-time data exchange and authentication. Ensure secure user authentication, image processing requests, and data retrieval from the backend. Testing & Optimization: Conduct rigorous UI/UX testing to ensure the application is intuitive, accessible, and bug-free. Optimize performance, responsiveness, and security for a smooth user experience. Deployment & Documentation: Deploy the web application in a test/staging environment for validation. Provide technical documentation on the front-end architecture, API connections, and deployment process.

Data analytics for rapid disease response

The goal of this project is to develop advanced disease modelling capacity in a livestock traceability system, that will allow for rapid disease response. The outcome is functionality that will allow authorities to track at-risk animals, premises and conveyance vehicles by accessing traceability datasets and applying data analytics based on disease parameters. The use case for this project will be Foot and Mouth Disease.

Deployment environment needed

The goal of this project is to create a deployment environment for a web application hosting an AI technology whose components (a full end-to-end pipeline) have been constructed, developed and well documented.