AI Photo and Video App Development
ICT & Open Learning/Innovation
Client company:WeSmile
Dante George
Hristo Tanchev
Nikola Manoilov
Project description
How can generative AI technologies be utilized to create innovative tools for photo and video generation that enhance user experiences at events?
This research question is broken down into several sub-questions, which the project aims to address:
- What are the current capabilities and limitations of generative AI in photo and video generation?
- How can user interfaces be designed to facilitate seamless interaction with AI-powered photo and video generation tools?
- What methods can be employed to ensure the quality and alignment of AI-generated content with user intent?
- How can the existing AI-powered photobooth product be improved and scaled to incorporate these new generative AI capabilities?
Context
The main project, undertaken in collaboration with WeSmile, is set within the context of rapidly evolving AI technologies and their application in enhancing user experiences at events. WeSmile, a leading provider of entertainment services for corporate events, has established a reputation for delivering memorable and engaging interactions through innovative solutions like their AI-powered photobooth. However, to maintain its competitive edge and pioneer status in the industry, WeSmile recognizes the need to continuously innovate and explore new trends.
Results
The WeSmile project, centered on advancing generative AI technologies for enhancing event experiences, culminated in several pivotal outcomes. These outcomes not only resulted in tangible products but also provided valuable insights that will drive future innovations in AI-driven event solutions.
Key Outcomes
AI-Powered Video Booth Prototype:
Product:
The primary tangible outcome of the project was the development of an AI-powered video booth prototype. This prototype leverages generative AI to create engaging and dynamic video content, tailored to enhance user experiences at events.
Value: The prototype showcases the practical application of AI in generating real-time video content, offering a novel and interactive feature for event attendees. This innovation positions WeSmile at the forefront of AI-driven entertainment, providing a unique selling point that differentiates their offerings from competitors.
Bias Mitigation in Generative Models:
Insight:
A significant portion of the project was dedicated to identifying and mitigating biases—specifically gender and racial biases—in generative AI models. Through extensive testing and model adjustments, the team successfully reduced these biases.
Value: Ensuring fairness and inclusivity in AI-generated content is crucial for maintaining ethical standards and user trust. This achievement not only enhances the quality of the generated content but also broadens its appeal to a more diverse audience, reinforcing WeSmile's commitment to inclusivity.
Project Pivot and Adaptability:
Insight:
Midway through the semester, the team pivoted to a new project with Wilrik, focused on generating music videos for his YouTube channel. This pivot allowed the team to apply their research in a practical context where generation time was less critical.
Value: The ability to pivot and adapt the project to meet new requirements demonstrated the team's flexibility and problem-solving capabilities. By successfully developing a comprehensive web application for video generation and Spotify canvas images, the team showcased their innovative approach and expanded the potential applications of their AI research.
Cloud Hosting and Scalability of the Photobooth:
Product:
The AI-powered photobooth was successfully transitioned to a cloud-hosted solution, enhancing its scalability and accessibility. The development included optimizing the backend to support cloud integration, ensuring seamless operation across different environments.
Value:
Hosting the photobooth in the cloud allows WeSmile to offer a more scalable solution that can handle larger volumes of user interactions and data processing. This enhancement significantly improves the product's robustness and flexibility, enabling it to be deployed at a wider range of events without compromising performance.
About the project group
The project group is a group of mix-semester developers keen on innovative and upcoming technology.