We have proposed an improved model for visual tracking using an adaptive coupled model. The model's advantage lays within the capability to track articulated objects using simple local features that are connected to a weak geometrical constellation. The model can robustly add and remove local features depending on probability maps of high-level image properties such as motion and color. The model also enables inclusion of additional probability maps based on arbitrary high-level image properties. We have analyzed the proposed tracker on a large video database and compared it with the current state-of-the-art. The experiments have shown that the proposed tracker outperforms the reference trackers based on multiple criteria. The IEEE Transactions on pattern analysis and machine intelligence journal is ranked in as the first of the 111 journals within the category EP - computer science, artificial intelligence. The journal has an impact factor of 4.908 and is one of the two most prestigious journals in the field of computer vision.
COBISS.SI-ID: 9431124
In this work we address the problem of binding – the ability to combine two or more modal representations of the same entity into a single shared representation, which is vital for every cognitive system operating in a complex environment. In order to successfully adapt to changes in a dynamic environment the binding mechanism has to be supplemented with cross-modal learning. We define the problems of high-level binding and cross-modal learning, model a binding mechanism in a Markov logic network and define its role in a cognitive architecture. This mechanism was also implemented in a multimodal heterogeneous integrated system for interactive learning in a dialogue with a tutor.
COBISS.SI-ID: 9172308
In this paper we propose a method for automatic assessment of aesthetic appeal of pho- tographs. We identify significant parameters that distinguish high quality photography from low quality snapshots. On the basis of these parameters, we defined calculable features for automatic assessment of photography aesthetics using machine learning methods. The calculation of features depends heavily on the identification of the subject in photographs. With the subject identified, we defined and implemented various features to analyze various aspects of a photograph. The features were tested on two datasets. First dataset was obtained from Flickr and manually labeled for evaluation. Second dataset was based on photographs from DPChallenge portal where subjects were identified with a face detection algorithm. Both experiments showed some promising results. In this article we specify the fea- tures which contribute to a successful classification of photographs, analyze their influence and discuss the results. In conclusion, we offer some suggestions for further research.
COBISS.SI-ID: 9441108
Over the recent years, low-level visual descriptors, among which the most popular is the histogram of oriented gradients (HOG), have shown excellent performance in object detection and categorization. We form a hypothesis that the low-level image descriptors can be improved by learning the statistically relevant edge structures from natural images. We validate this hypothesis by introducing a new descriptor called the histogram of compositions (HoC). HoC exploits a learnt vocabulary of parts from a state-of-the-art hierarchical compositional model. Furthermore, we show that HoC is a complementary HoC descriptor to HOG. We experimentally compare our descriptor to the popular HOG descriptor on the task of object categorization. We have observed approximately 4% improved categorization performance of HoC over HOG at lower dimensionality of the descriptor. Furthermore, in comparison to HOG, we show a categorization improvement of approximately 10% when combining HOG with the proposed HoC.
COBISS.SI-ID: 9671508
For the successful operation in real-world environments, a mobile robot requires an effective spatial model. In this work we introduced a new compositional hierarchical representation of space that is based on learning of statistically significant observations, in terms of frequency of occurrence of various shapes in the environment. We also proposed a new low-level image descriptor, by which we demonstrated the performance of our representation in the context of room categorization problem.
COBISS.SI-ID: 9674068