Institut für Raumbezogene Informations- und Messtechnik
Hochschule Mainz - University of Applied Sciences

Knowledge-based object Detection in Image and Point cloud (KnowDIP)

The KnowDIP project aims at the conception of a framework for an automatic object detection in unstructured and heterogeneous data. This framework uses a representation of human knowledge in order to improve the flexibility, the accuracy, and the efficiency of the processing.
Motivation und Ziele: 

The KnowDPI project is to create a framework for an adaptive object detection on different types of data. For that, the framework will use the semantic technologies in order to understand the context of the detection and determine the most adapted algorithms for the processing. This processing will be created according to the type of user query, and available algorithms for this type and quality of data in order to obtain the best result. In order to realize a such framework, three main goals must be attempted. The first is the design of an automatic and adaptive processing of heterogeneous data. The second is to optimize dynamically the algorithms combination and parameterization. The third will be to create a system of learning which could improve the object detection.


The first year of this project has been composed of two main activities. The first activities has been to research in the literature if some existing system of object detection use the semantic technologies in order to identify the different existing approaches. The second activity which corresponds to design the framework model, begin quickly. This first step of design has conducted to the identification of the framework requirements. The conceptual model of the framework is composed of four modules which will interact together in order to satisfied the project goal. Each module has a specific role. One will content all information about data, algorithms and objects. Another will manage the algorithm execution according to the decision made by the third module. Finally, the fourth module will be a learning system.


  The result of these activities is a set of concepts which will be the base for the development of this framework. Some of these concepts are illustrated by following pictures.  


Zeitraum:     01.05.2016 - 30.09.2019
  • a) 3D Point Cloud, b) Object detection


Mit zwei Vorträgen war das i3mainz auf den 18. Oldenburger 3D-Tagen vom 6. bis 7. Februar 2018 vertreten.

Jean-Jacques Ponciano stellte einige…

Verwandte Projekte

Im Zuge der Publikation und Analyse der 3D-Daten eines unberührten Gruftkomplexes zu Qatna ist die Bereitstellung der Daten anhand einer plattformunabhängigen Web-Applikation besch…
Zur Publikation und Unterstützung der archäologischen Interpretation sollte 2016 eine plattformunabhängige Applikation entwickelt werden mit dessen Hilfe die unterschiedlichen 3D-D…


Knowledge-based object recognition in point clouds and image data sets


J.J. Ponciano,
A. Trémeau