Project Overview

Pioneering the frontier of autonomous environmental monitoring with state-of-the-art AI.

Project Title: BA Thesis: ISEEU-Net - Real-Time AI-Powered Change Detection for Drones
Duration: 6 Months
Role: Developer & Researcher
Technologies Used: Python, Keras, U-Net, Nvidia Jetson NX, Unity, Real Virtuality 4

Problem Statement and Objectives

Problem Description: The demand for advanced change detection in observed areas using integrated camera systems on drones, hindered by the scarcity of public datasets for training robust neural networks.
Project Objectives: To create a high-performance change detection system, capable of operating in real time with drones, trained on a synthesis of HD datasets and synthetic data generated using game engines.
Target Audience/Market: Tech startups, environmental monitoring agencies, and AI research communities.

Key Challenges: Generating a synthetic dataset representative of real-world scenarios and adapting neural network architecture to meet real-time processing requirements.
Solutions Developed: Created a hybrid dataset and trained a modified U-Net architecture capable of seamless performance on consumer-grade hardware.
Impact of Solutions: Achieved a system on par with current technologies, allowing for movement detection in various conditions without heavy computational demands.

Lifecycle Overview: From dataset generation and architecture design to training, optimizing, and evaluating against industry benchmarks, the process was iterative and rigorous.
Phases of Development: Dataset creation, neural network training, system evaluation, and optimization.
Collaboration: Worked closely with RWU and HATtec GmbH for the successful completion of the project, utilizing their resources and expertise.

Milestones: Developed ISEEU-Net, which can detect changes during omnidirectional flight, accommodate lighting changes, and identify small object movements.
Final Outcomes: Real-time, reliable change detection model which excels on standard benchmarks and operates effectively on the Nvidia Jetson NX.
Personal Learning: Mastered synthetic data integration and real-time AI deployment, paving my way in AI-driven environmental technology.

Screenshots/Diagrams: [Links or placeholders for visual representations of the project #todo]
Live Demos/Repositories: [Provide links to demos, repositories, or project documentation here #todo]

Project Impact: The ISEEU-Net project signals a significant advancement in drone technology applications for real-time environmental monitoring, carving out new possibilities for industry practices.
Career Reflection: This project not only solidified my technical expertise but also my commitment to harnessing AI for transformative solutions in tech—a stepping stone in my entrepreneurial journey.