ROSSINI: Reconstructing 3D structure from single images: a perceptual reconstruction approach

Lead Research Organisation: University of Surrey
Department Name: Vision Speech and Signal Proc CVSSP

Abstract

Abstracts are not currently available in GtR for all funded research. This is normally because the abstract was not required at the time of proposal submission, but may be because it included sensitive information such as personal details.
 
Description ROSSINI has developed a test bed and protocols of large scale evaluation of monodepth algorithms. It has ran three international challenges using this data and code which have been released to the community. The legacy of which continues. The project has also ran tutorials and developed state of the art approaches to depth estimation from single images.
Exploitation Route The code and datasets have been released to the community and we have ran three challenges making use of them. These challenges are continuing with the momentum we created in the project. A new project is building on some of the IP generated and we are currently discussing licensing options.
Sectors Aerospace

Defence and Marine

Creative Economy

Digital/Communication/Information Technologies (including Software)

Transport
 
Description Several industrial collaborations have now built on the IP and knowledge generated in this project. There is potential for licensing in the future.
First Year Of Impact 2023
Sector Aerospace, Defence and Marine,Agriculture, Food and Drink,Digital/Communication/Information Technologies (including Software),Transport
 
Description Aston University 
Organisation Aston University
Department Department of Psychology
Country United Kingdom 
Sector Academic/University 
PI Contribution Collaborative research as part of ROSSINI project with Dr Andrew Schofield at Aston University
Collaborator Contribution Collaborative, multidisciplinary research project
Impact Not yet, ongoing
Start Year 2019
 
Description Southampton University 
Organisation University of Southampton
Country United Kingdom 
Sector Academic/University 
PI Contribution Collaborative research as part of ROSSINI project with Prof Wendy Adams at Southampton University
Collaborator Contribution Collaborative, multidisciplinary research project
Impact Not yet, ongoing
Start Year 2019
 
Description York University Canada 
Organisation York University Toronto
Country Canada 
Sector Academic/University 
PI Contribution Collaborative research as part of ROSSINI project with Prof James Elder at York University Canada
Collaborator Contribution Direct input into program of research at both an advisory and technical level
Impact Not yet, ongoing
Start Year 2019
 
Description One Day BMVA Symposium: 3D worlds from 2D images in humans and machines. 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Professional Practitioners
Results and Impact When humans view a photograph they perceive the 3D world that constructed the image. They can, for example, describe the depth relationships between objects, plan a route through the scene and imagine the scene from a different viewpoint. This process is automatic and compulsive. For example, even though humans possess size constancy they will readily misinterpret the size of a person in order to make sense of the rest of the scene as a 3D world. State of the art computer vision systems are now also very good at reconstructing 3D layout from 2D images (3D uplift) although, unlike humans, this is often restricted to specific domains or requires multiple views. This workshop will consider recent developments in 3D uplift as well as our current knowledge of scene understanding in human vision.
Year(s) Of Engagement Activity 2020
URL https://britishmachinevisionassociation.github.io/meetings