Intrinsic research team, in collaboration with Texas A&M University, published a new paper on arXiv demonstrating that stereo vision does not require specific architectural inductive biases and developed a system achieving higher quality stereo 3D reconstruction with much faster inference.
Public source
Publisher name
Public post
How can robots and AI systems better visualize the world in 3D? 🌍 Stereo vision is a technique that uses two camera views to perceive depth. For years, researchers and…
Company
Intrinsic
- Industry
- Software Development
- Location
- Mountain View, US
- Company size
- 201–500 employees
About Intrinsic
Intrinsic is an AI robotics group at Google. Our mission is to democratize intelligent robotics for millions more businesses, entrepreneurs, and developers. We are a multidisciplinary team of engineers, roboticists, designers, and technologists who are passionate about unlocking the creative and economic potential of industrial robotics.
See moreLatest activity
Latest activity from Intrinsic
3 signals
Presence & Recognition
Intrinsic is heading to the International Manufacturing Technology Show in Chicago, IMTS, with a booth at Booth 236233 in the North Building, Level 3 from September 14 to 19.
Presence & Recognition
Intrinsic is displaying its platform at the International Machine Tools and Suppliers Conference (IMTS).
Discover more
Similar signals
Similar public activity from other companies.
Research & Knowledge
Simulon
Simulon published the work of NBS: No Bias Stereo, an end-to-end Vision Transformer model for stereo matching that removes architectural inductive biases.
Research & Knowledge
Tristar AI
Tristar AI is exploring how computer vision can track individual parts throughout a production line without printed labels, applied tags, or additional consumables using visual embeddings.
Research & Knowledge
Adobe
Adobe published a new paper titled LumiTokens exploring latent, neural-ish representations for scene encoding in image and video generation.
Research & Knowledge
Perciv AI
Perciv AI's paper 4D-RaDiff: Latent Diffusion for 4D Radar Point Cloud Generation has been accepted at the ECCV 2026 DriveX Workshop.
Research & Knowledge
ARGO Vision