Marko Boben

Similarity-based cross-layered hierarchical representation for object categorization

This paper proposes a new concept in hierarchical representations that exploits features of different granularity and specificity coming from all layers of the hierarchy. The concept is realized within a cross-layered compositional representation learned from the visual data. We show how similarity connections among discrete labels within and across hierarchical layers can be established in order to produce a set of layer-independent shape-terminals, i.e. shapinals. We thus break the traditional notion of hierarchies and show how the category-specific layers can make use of all the necessary features stemming from all hierarchical layers. This, on the one hand, brings higher generalization into the representation, yet on the other hand, it also encodes the notion of scales directly into the hierarchy, thus enabling a multi-scale representation of object categories. By focusing on shape information only, the approach is tested on the Caltech 101 dataset demonstrating good performance in comparison with other state-of-the-art methods.

Object Categorization: Learning Hierarchical Compositional Representations of Object Structure

Visual categorization of objects has captured the attention of the vision community for decades [10]. The increased popularity of the problem witnessed in the recent years and the advent of powerful computer hardware have led to a seeming success of categorization approaches on the standard datasets such as Caltech 101 [15]. However, the high discrepancy between the accuracy of object classification and detection/segmentation [14] suggests that the problem still poses a significant and open challenge. The recent preoccupation with tuning the approaches to specific datasets might have precluded the attention from the most crucial issue: the representation [41]. This paper will focus on what we believe are two central representational design principles, namely a hierarchical organization of categorical representations, more specifically, the principle of hierarchical compositionality, and statistical, bottom-up learning. Given images of complex scenes, objects must be inferred from the pixel information through some recognition process. This requires an efficient and robust matching of the internal object representation against the representation produced from the scene. Despite the seemingly effortless performance of human perception, the diversity and the shear number of visual object classes appearing in various scales, 3D positions and articulations, which additionally interact with each other (occlusion, clutter, etc.), have placed a great obstacle to the task. In fact, it has been shown by Tsotsos in 1990 [53] that the unbounded visual search is NP complete and thus approximate, hierarchical solutions might be the most promising/plausible way to tackle the problem. This line of architecture is also consistent with the findings on biological systems [44, 9]. A number of authors have further emphasized these computational considerations [13, 23, 2, 47, 26], suggesting that matching should be performed at multiple hierarchical stages, in order to gradually and coherently limit the otherwise computationally prohibitive search space [13, 53, 8, 3, 33, 2, 23, 17, 47, 19]. While hierarchies presented a natural way to represent objects in the early vision works [13, 24, 32, 11], surprisingly, they have not become an integral part of the modern vision approaches. Hierarchical representations can derive and organize the features at multiple levels that build on top of each other by exploiting the shareability of features among more complex compositions or objects themselves [13, 7, 21, 2,

The 10-cages and derived configurations

Symmetry properties of the three 10-cages on 70 vertices are investigated. Being bipartite, these graphs are Levi graphs of triangle- and quadrangle-free (353) configurations. For each of these graphs a Hamilton cycle is given via the associated LCF notation. Furthermore, the automorphism groups of respective orders 80, 120, and 24 are computed. A special emphasis is given to the Balaban 10-cage, the first known example of a 10-cage (Rev. Roumaine Math. Pure Appl. 18 (1973) 1033-1043), and the corresponding Balaban configuration. It is shown that the latter is linear, that is, it can be realized as a geometric configuration of points and lines in the Euclidean plane. Finally, based on the Balaban configuration, an infinite series of linear triangle-free and quadrangle-free ((7n)3) configurations is produced for each odd integer n?5.

Real-time coarse-to-fine topologically preserving segmentation

In this paper, we tackle the problem of unsupervised segmentation in the form of superpixels. Our main emphasis is on speed and accuracy. We build on [31] to define the problem as a boundary and topology preserving Markov random field. We propose a coarse to fine optimization technique that speeds up inference in terms of the number of updates by an order of magnitude. Our approach is shown to outperform [31] while employing a single iteration. We evaluate and compare our approach to state-of-the-art superpixel algorithms on the BSD and KITTI benchmarks. Our approach significantly outperforms the baselines in the segmentation metrics and achieves the lowest error on the stereo task.

A coarse-to-fine taxonomy of constellations for fast multi-class object detection

In order for recognition systems to scale to a larger number of object categories building visual class taxonomies is important to achieve running times logarithmic in the number of classes [1,2]. In this paper we propose a novel approach for speeding up recognition times of multi-class part-based object representations. The main idea is to construct a taxonomy of constellation models cascaded from coarse-to-fine resolution and use it in recognition with an efficient search strategy. The taxonomy is built automatically in a way to minimize the number of expected computations during recognition by optimizing the cost-to-power ratio [3]. The structure and the depth of the taxonomy is not pre-determined but is inferred from the data. The approach is utilized on the hierarchy-of-parts model [4] achieving efficiency in both, the representation of the structure of objects as well as in the number of modeled object classes. We achieve speed-up even for a small number of object classes on the ETHZ and TUD dataset. On a larger scale, our approach achieves detection time that is logarithmic in the number of classes.

Small Triangle-Free Configurations of Points and Lines

AbstractIn the paper we show that all combinatorial triangle-free configurations for v ≤ 18 are geometrically realizable. We also show that there is a unique smallest astral (183) triangle-free configuration and its Levi graph is the generalized Petersen graph G(18,5). In addition, we present geometric realizations of the unique flag transitive triangle-free configuration (203) and the unique point transitive triangle-free configuration (213).

Optimization Framework for Learning a Hierarchical Shape Vocabulary for Object Class Detection.

This paper proposes a stochastic optimization framework for unsupervised learning of a hierarchical vocabulary of object shape intended for object class detection. We build on the approach by [6], which has two drawbacks: 1.) learning is performed strictly bottom-up; and 2.) the selection of vocabulary shapes is done solely on their frequency of appearance. This makes the method prone to overfitting of certain parts of object shape while losing the more discriminative shape information. The idea of this paper is to cast the vocabulary learning into an optimization framework that iteratively improves the hierarchy as a whole. Optimization is two-fold: one that learns and selects the vocabulary of shapes at each layer in a bottom-up phase and the other that extends/improves it by top-down feedback from the higher layers. The algorithm then loops between the two learning stages several times. We have evaluated the proposed learning approach for object class detection on 11 diverse object classes taken from the standard recognition data sets. Compared to the original approach [6], we obtain a 3 times more compact vocabulary, a 2:5 times faster inference, and a 10% higher detection performance at the expense of 5 times longer training time (25min vs 5min). The approach attains a competitive detection performance with respect to the current state-of-the-art at both, faster inference as well as shorter training times.

Adding discriminative power to a generative hierarchical compositional model using histograms of compositions

We highlight the problem of poor discriminative power in hierarchical compositions.We combine generative hierarchical model with extracted discriminative features.We propose histogram of compositions (HoC) to capture discriminative features.HoC descriptor reduces similar category misclassification and phantom detections.Compared to HOG descriptor HoC classifier performs better in most cases. In this paper we identify two types of problems with excessive feature sharing and the lack of discriminative learning in hierarchical compositional models: (a) similar category misclassifications and (b) phantom detections in background objects. We propose to overcome those issues by fully utilizing a discriminative features already present in the generative models of hierarchical compositions. We introduce descriptor called histogram of compositions to capture the information important for improving discriminative power and use it with a classifier to learn distinctive features important for successful discrimination. The generative model of hierarchical compositions is combined with the discriminative descriptor by performing hypothesis verification of detections produced by the hierarchical compositional model. We evaluate proposed descriptor on five datasets and show to improve the misclassification rate between similar categories as well as the misclassification rate of phantom detections on backgrounds. Additionally, we compare our approach against a state-of-the-art convolutional neural network and show to outperform it under significant occlusions.

Adding Discriminative Power to Hierarchical Compositional Models for Object Class Detection

In recent years, hierarchical compositional models have been shown to possess many appealing properties for the object class detection such as coping with potentially large number of object categories. The reason is that they encode categories by hierarchical vocabularies of parts which are shared among the categories. On the downside, the sharing and purely reconstructive nature causes problems when categorizing visually-similar categories and separating them from the background. In this paper we propose a novel approach that preserves the appealing properties of the generative hierarchical models, while at the same time improves their discrimination properties. We achieve this by introducing a network of discriminative nodes on top of the existing generative hierarchy. The discriminative nodes are sparse linear combinations of activated generative parts. We show in the experiments that the discriminative nodes consistently improve a state-of-the-art hierarchical compositional model. Results show that our approach considers only a fraction of all nodes in the vocabulary (less than 10%) which also makes the system computationally efficient.

Room Categorization Based on a Hierarchical Representation of Space

For successful operation in real-world environments, a mobile robot requires an effective spatial model. The model should be compact, should possess large expressive power and should scale well with respect to the number of modelled categories. In this paper we propose a new compositional hierarchical representation of space that is based on learning statistically significant observations, in terms of the frequency of occurrence of various shapes in the environment. We have focused on a two-dimensional space, since many robots perceive their surroundings in two dimensions with the use of a laser range finder or sonar. We also propose a new low-level image descriptor, by which we demonstrate the performance of our representation in the context of a room categorization problem. Using only the lower layers of the hierarchy, we obtain state-of-the-art categorization results in two different experimental scenarios. We also present a large, freely available, dataset, which is intended for room categorization experiments based on data obtained with a laser range finder.