Objectives

We propose a new solution for pixel-level semi-automatic annotation of large datasets, without constraints based on application or number of identified classes. Our aim is to develop a very reliable marking and querying workflow which addresses the problems related to big data paradigms. There are three major tasks which need to be accounted for, namely the Image Database Server, the Marking and Querying Tool and the Database Administration and Management System.

The projects’ main goals are strongly related to the objectives of the 2015-2020 PNIII component part of the Research, Development and Innovation National Plan. They will focus on bringing added value to the area of Informational Technology and Communications domain, by working on completing the following objectives:

  • The increase of the competition capabilities on international level of Romanian research teams, by applying fundamental research concepts in maturing the technological development; the project will contribute to the knowledge increment by developing a fully validated semi-automatic intelligent marking tool;
  • The integration, testing and validation of the semi-automatic marking and querying tool in a complete pixel-level annotation workflow;
  • Publishing of at least 2 scientific articles in peer-reviewed publications; (e.g. IEEE Transactions, IEEE Letters, Signal Processing Magazine) or at a IEEE Conference;
  • The completion of a fully developed, tested and validated product, in the form of the Marking and Querying Tool;
  • The generation of a complete service in the form of a fully annotated dataset with a minimum number of 100.000 images, acquired in 2 different scenarios (indoor, outdoor) and containing 30 different classes (15 for each scenario).