University of Cambridge > Talks.cam > Technical Talks - Department of Computer Science and Technology  > Roku: Beyond Generative AI: Achieving Complete Control in Synthesising Testing Data with Unreal Engine

Roku: Beyond Generative AI: Achieving Complete Control in Synthesising Testing Data with Unreal Engine

Add to your list(s) Download to your calendar using vCal

If you have a question about this talk, please contact Ben Karniely.

Whilst generative AI has advanced significantly in recent years, when it comes to accountability and assurance of AI systems, the challenge of achieving precise control in synthesizing testing data for algorithm development remains significant. This session will provide an exciting opportunity to uncover how thorough tests are done in the industry before we can deploy AI/CV algorithms in real-world applications! We will delve into:

• Unreal Engine’s advanced capabilities for precise parameterization and simulation of nuanced real-world scenarios, such as subtle lighting changes, tree swaying, and controlled object movements, which ensures a higher degree of realism and reproducibility in testing environments.

• Generation of testing videos tailored for IoT camera algorithms, with a specific focus on motion detection, motion zone suggestion, and event detection.

• Insights from testing – Understanding what tests reveal about the performance and reliability of algorithms, enabling developers to refine and optimize AI models for real-world challenges

Also in attendance will be members of Roku’s Early Career Recruiting team to discuss our current internship and graduate opportunities.

Attendees can sign up via the below link: https://app.ripplematch.com/t/f93dba87

Some catering will be provided following the talk

This talk is part of the Technical Talks - Department of Computer Science and Technology series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.

 

© 2006-2025 Talks.cam, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity