SagivTech’s technology is based on state-of-the-art research conducted in the fields of computer vision, machine learning, deep learning and efficient computing.
We are fortunate to have been chosen by the European Commission to take part in and even lead several prestigious international projects in the field of machine vision. These include: SceneNet, UnLocX and 3D MASSOMICS.
SceneNet was a Future Emerging Technologies (FET) research project funded in the framework of the FET SME grant of the European Commission. SagivTech was the coordinator of SceneNet with the research partners University of Bremen and Ecole Polytechnique Federale de Lausanne (EPFL). The aim of SceneNet was to use the power of crowd sourcing, in the form of multiple mobile phone users, to create a higher quality 3D video scene experience that can be shared via social networks. The backbone of SceneNet is advanced technology for video streaming, image registration, 3D reconstruction, visualization and GPU computing. The individual videos presenting multiple viewpoints of the same scene are combined to create an enhanced 3D video experience. A typical SceneNet scenario can be a rock concert, a sports event, a wedding ceremony, breaking news events and any other multiple mobile users’ crowded event.
The SceneNet pipeline starts at the mobile devices where the video streams are acquired, pre-processed and transmitted along with a tagging identity to the server. At the server, the various video streams are registered and synchronized and then submitted to 3D reconstruction to generate a multi-view video scene that can be edited and shared by the users using a 3D/4D visualizer.
The SceneNet pipeline requires immense computing power on both the mobile device and the cloud server side.
The technological research conducted in SceneNet serves as the infrastructure for a joint event documentation application that can be appealing to users. More information on SceneNet can be found here.
SceneNet was rated “excellent” at its completion and has since received widespread recognition for its contribution to crowd sourcing based 3D reconstruction of video.
• Cordis publication
• ISERD publication on the EU program
• Times of Israel
• Newsroom of the Digital Agenda for Europe
• Publication in LaserFocusWorld
• Publication in Jewish Business News
• Article of Women@GTC Focus on Innovation, Inspiration and Roadmap for Inclusion featuring, among others, Chen Sagiv – See more here
UnLocX was a Future Emerging Technologies (FET) project that dealt with uncertainty principles and function systems for efficient coding schemes. UnLocX brought together mathematicians, algorithms developers and code optimization experts. UnLocX was rooted in the theory of uncertainty principles and aimed at developing a framework for constructing problem adapted, ultra-efficient algorithms concerning (de-)coding and analyzing/synthesizing signals and images.
In the framework of UnLocX, SagivTech was responsible for the advanced GPU implementations of these algorithms. More information can be found in the UnLocX website
3D MASSOMICS was a research project funded by the European Commission in the EU FP7 HEALTH Programme. The goals of the project were to introduce statistical methods for studying variations of Mass Spectrometry (MS) data in order to develop reproducible data acquisition protocols and to develop and evaluate statistical methods for un-supervised and supervised statistical analysis of 3D MS data. In the scope of 3D-MASSOMICS, SagivTech led two main activities: the first dealt with efficient GPU implementation of supervised and un-supervised methods. This activity resulted in a GPU library with efficient implementations for the Measure of Chaos, Hierarchical Clustering, PCA using SVD, PLSA, Median filter, NIPLAS and SVM. The second activity led by SagivTech, involved research on the application of deep learning techniques to MS images for the use case of spectra selection. More information on the 3D-MASSOMICS project can be found here.