Top-Down Visual Attention with Complex Templates

摘要

GET Lab, University of Paderborn, Pohlweg 47-49, 33098 Paderborn, Germanyftuennermann, born, mertschingg@get.upb.deKeywords: Visual Attention, Saliency, Top-Down Control, Visual Search, Object Detection.Abstract: Visual attention can support autonomous robots in visual tasks by assigning resources to relevant portions ofan image. In this biologically inspired concept, conspicuous elements of the image are typically determinedwith regard to different features such as color, intensity or orientation. The assessment of human visual atten-tion suggests that these bottom-up processes are complemented – and in many cases overruled – by top-downinfluences that modulate the attentional focus with respect to the current task or a priori knowledge. In arti-ficial attention, one branch of research investigates visual search for a given object within a scene by the useof top-down attention. Current models require extensive training for a specific target or are limited to verysimple templates. Here we propose a multi-region template model that can direct the attentional focus withrespect to complex target appearances without any training. The template can be adaptively adjusted to com-pensate gradual changes of the object’s appearance. Furthermore, the model is integrated with the frameworkof region-based attention and can be combined with bottom-up saliency mechanisms. Our experimental re-sults show that the proposed method outperforms an approach that uses single-region templates and performsequally well as state-of-the-art feature fusion approaches that require extensive training.