Project Overview

DanOffice partnered with ISE students to enhance their image recognition capabilities through an advanced AI model retraining. This project focused on improving the accuracy and performance of an existing computer vision system. The goal is to allow on-site, non-technical individuals to retrain and publish an AI model in a line clearance process.

Key Achievements

1. Model Improvement: Achieved improvements in image classification
2. Process Improvement: Non-technical people can retrain and publish an AI model in a line clearance process

Technical Overview

The project utilized state-of-the-art tech, including:

1. AI-codebase is run on a Jupyter notebook
2. Google Colab, Cloud-based Python code execution with GPU support on Google’s GPU servers
3. Google Drive, for images and file storage
4. HTML and CSS for the user interface
5. Docker, App containerization
6. Azure, as a cloud hosting service
7. ASP.NET for the Backend
8. Identity framework for user authentication
9. Entity Framework Core for database interactions
10. Database: SQLite server

Impact

The enhanced AI image recognition has improved DanOffice’s product offering, enabling non-technical individuals to retrain and publish AI models.