Project Website: ---


Start Date:
July of 2016

End Date:
June of 2018

Selectively Accessing Light Field Faces over Information Centric Networking (SeLF-ICN)

The Internet, as we know it, is increasingly being integrated into several different fields of our society. As such, new utilization and performance requirements are being placed over a communications architecture, whose origin dates for several decades. In its conception, scenarios such as ultra-high definition video streaming, or mobile terminals moving between different wireless technologies, were not considered and were only later on added-in to systems that used the protocol.

In this way, new lines of thinking have started to be taken, considering whole new Internet architectures that explore radical new approaches, more tailored to the ways we use on-line services and devices today. One of such examples are Information Centric Networks (ICN) . In this kind of environment, the network layer operates as a natural extension of how we request information, calling it by its name, and intrinsically supporting important mechanisms (such as caching and mobility support) out-of-the-box. This concept has generated a lot of interest from an academia, industry and standardization point of view, having conceived a large number of different international research projects that exploited these concepts towards different ends (i.e., FP7 4WARD, FP7 SAIL, FP7 PURSUIT, H2020 POINT, NSF Named Data Networking, etc.). From these projects, several proof of concept experimental testbeds and software have surfaced as well, such as Blackhadder, CCNx and NDN, allowing for the exploration of ICN capabilities in different scenarios.

One aspect that hasn't yet been explored is the capability of name-based mechanisms (such as the ones used by ICN approaches such as CCNx and NDN) in potentiating new services in the multimedia sphere. One of such possibilities is using these name-based mechanisms, along with the other intrinsic ICN capabilities, to empower video surveillance scenarios.

In this context, the SeLF-ICN project aims to develop a new visual technology where certain elements belonging to light field images (e.g., the faces of persons captured by a security camera) can be extracted and identified, and integrate that technology with a new mechanism for referencing the extracted images by a name that is routable directly in the network layer (instead of using a server address to locate a video file therein, as in current IP). The project proposes to do so by exploiting the richer information that is captured by the emerging light field cameras whose scene representation allows, for instance, to extract objects at multiple focus planes independently of the focus used when taking the picture, to analyze the image from different viewpoints, exploiting the available directional information, and finally to extract richer feature sets for various analysis purposes. In the proposed approach, the name-based operational mechanisms of ICN, will be used to name and reach extracted elements from the video content. In this sense, the project will enable a network to become a large video database, with the same name-based networking request mechanisms to be used to request specific elements present inside the video images, such as faces from people.

SeLF-ICN is a joint research project between the Aveiro and Lisbon poles of the Instituto de Telcomunicações, involving the "Telecommunications and Networking" (ATNoG) and the "Multimedia and Signal Processing" groups, respectively.

The project has started on July 2016, going all the way to June 2018.

The team

IT Aveiro: Daniel Corujo, Rui L. Aguiar
IT Lisboa: Paulo Correia, Fernando Pereira

The Tasks

Task 1: Integrated Project Management
Task 2: Requirements and Conceptual Design
Task 3: Development of Content Identification in Content-Centric

Task 4: Light Field Face Analysis
Task 5: Integration of name-based network mechanisms with light field image analysis
Task 5: Integration of name-based network mechanisms with light field image analysis