The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence) Computer Vision for Remote AR with Flora Tasse - #390
Jul 9, 2020
Flora Tasse, Head of Computer Vision & AI Research at Streem, shares her journey from Cameroon to becoming an expert in AR/VR and 3D mesh environments. She discusses the innovative blend of 2D image processing with natural language for 3D modeling and the complexities of remote AR, including real-time 3D meshing challenges. Flora also explores six degrees of freedom in object tracking and highlights the exciting possibilities of AR technology in enhancing consumer experiences and user interactions.
AI Snips
Chapters
Transcript
Episode notes
Flora's Journey into Computer Vision
- Flora Tasse, driven by a childhood fascination with Jurassic Park's special effects, pursued computer vision.
- Her journey led her from Cameroon to South Africa, Cambridge, and finally to founding Selerio.
NLP Improves 2D to 3D Conversion
- Tasse's PhD research revealed that incorporating NLP significantly improves 2D to 3D model creation.
- Semantic features, derived from language information, enhance the accuracy of the generated 3D models.
Remote AR's Unique Challenges
- Remote AR differs significantly from traditional AR, as it involves someone remotely experiencing another person's view.
- This presents unique challenges that computer vision helps address, such as meshing the remote environment.