Research objectives and content
This research proposal aims at understanding human object recognition within an inter-disciplinary framework, involving an interplay between connectionist models and experimental work. Current models of object recognition mostly lack the following properties: top-down influence on object recognition, explanation to a broad range of data, processing of real world images and consideration of attentional effects On object recognition. In previous work a Selective Attention Identification Model (SAIM) was developed which fulfils some of these critics. This research project intends to extend SAIM to SAIM2.0 to meet all of these requirements. The principle idea of SAIM2.0 is a minimisation of an energy function and a graph representation of objects which proved to be capable of processing real world images. The accompanying experiments and resulting explanations of SAIM2.0 will include attentional effects and the influence of knowledge on object recognition and grouping.
Training content (obJective, benefit and expected impact)
The proposed project will increase the already existing experience and skills in Computer Vision by the application of new methods, e.g. graph representation of objects. In addition it will lead to new experience and skills in Cognitive Psychology within an international collaboration with excellent scientists, such as Professor Humphreys and his colleagues of the Cognitive Science Research Centre at the University of Birmingham. Concluding, the training will lead to a very good foundation of an academic career in the field of Cognitive Science.
Links with industry / industrial rele\ ulce (22)
On the long term the results of this project can be applied in industrial area as well, e.g. vision based automated production.