Web-based computer vision sandbox

Research areas

Description

The computer vision community currently produces increasingly impressive results, solving challenging problems such as semantic scene understanding using deep neural networks, or accurate dense reconstructions of dynamic 3D environments. On one hand, the underlying mathematical techniques are often quite involved, which makes it difficult for non-expert users to reproduce results or judge the suitability of a particular method. On the other hand, published code often requires additional dependencies, or again expert knowledge to be used. The community is to a certain degree missing a user-friendly platform where methods can be accessed and applied to user-specific data, without the requirement of any expert knowledge in mathematics, computer vision, or software engineering. The goal of this project will be the conceptual design and prototype integration of such a platform. The idea is to leverage the power of the internet, and envisage a web-portal where users can upload data and play around with various problem solvers. The concept potentially leads to an easier dissemination of scientific results, and a more conceivable discovery of future commercial opportunities. type of work: implementational

Goals

-Research on existing frameworks for the design of interactive webpages -Familiarization with computer vision toolboxes such as OpenCV -Design and implementation of a prototype that allows easy access and execution of several, basic computer vision algori

Requirements

-Modern web programming -C++ coding skills may be an advantage -Time, motivation, and persistence! -Being prepared for a challenging project!

Gain

The student will have the opportunity to be part of a dynamic team of active, internationally renowned researchers in 3D computer vision, and contribute to the development of cutting-edge research demonstrators. The student will also learn how to design, implement, benchmark, and maintain software frameworks in industry standard, and thus work towards the development of a framework with high market potential.

Updated:  8 September 2015/Responsible Officer:  Dean, CECS/Page Contact:  CECS Marketing