Bag-of-Features Approach to Unsupervised Visual Object Categorisation

The large and growing number of digital images is making manual image search laborious. Only a fraction of the images contain metadata that can be used to search for a particular type of image. Thus, the main research question of this thesis is whether it is possible to learn visual object categories directly from images. Computers process images as long lists of pixels that do not have a clear connection to high-level semantics which could be used in the image search. There are various methods introduced in the literature to extract low-level image features and also approaches to connect these low-level features with high-level semantics. One of these approaches is called Bag-of-Features which is studied in the thesis. In the Bag-of-Features approach, the images are described using a visual codebook. The codebook is built from the descriptions of the image patches using clustering. The images are described by matching descriptions of image patches with the visual codebook and computing the number of matches for each code. In this thesis, unsupervised visual object categorisation using the Bag-of-Features approach is studied. The goal is to find groups of similar images, e.g., images that contain an object from the same category. The standard Bag-of-Features approach is improved by using spatial information and visual saliency. It was found that the performance of the visual object categorisation can be improved by using spatial information of local features to verify the matches. However, this process is computationally heavy, and thus, the number of images must be limited in the spatial matching, for example, by using the Bag-of-Features method as in this study. Different approaches for saliency detection are studied and a new method based on the Hessian-Affine local feature detector is proposed. The new method achieves comparable results with current state-of-the-art. The visual object categorisation performance was improved by using foreground segmentation based on saliency information, especially when the background could be considered as clutter.

Unsupervised visual object categorization

The number of digital images has been increasing exponentially in the last few years. People have problems managing their image collections and finding a specific image. An automatic image categorization system could help them to manage images and find specific images. In this thesis, an unsupervised visual object categorization system was implemented to categorize a set of unknown images. The system is unsupervised, and hence, it does not need known images to train the system which needs to be manually obtained. Therefore, the number of possible categories and images can be huge. The system implemented in the thesis extracts local features from the images. These local features are used to build a codebook. The local features and the codebook are then used to generate a feature vector for an image. Images are categorized based on the feature vectors. The system is able to categorize any given set of images based on the visual appearance of the images. Images that have similar image regions are grouped together in the same category. Thus, for example, images which contain cars are assigned to the same cluster. The unsupervised visual object categorization system can be used in many situations, e.g., in an Internet search engine. The system can categorize images for a user, and the user can then easily find a specific type of image.

Comparison of Local Feature Detectors and Descriptors for Visual Object Categorization

Local features are used in many computer vision tasks including visual object categorization, content-based image retrieval and object recognition to mention a few. Local features are points, blobs or regions in images that are extracted using a local feature detector. To make use of extracted local features the localized interest points are described using a local feature descriptor. A descriptor histogram vector is a compact representation of an image and can be used for searching and matching images in databases. In this thesis the performance of local feature detectors and descriptors is evaluated for object class detection task. Features are extracted from image samples belonging to several object classes. Matching features are then searched using random image pairs of a same class. The goal of this thesis is to find out what are the best detector and descriptor methods for such task in terms of detector repeatability and descriptor matching rate.

An Evaluation of Local Feature Combiners for Robot Visual Localization

In the last decade, local image features have been widely used in robot visual localization. To assess image similarity, a strategy exploiting these features compares raw descriptors extracted from the current image to those in the models of places. This paper addresses the ensuing step in this process, where a combining function must be used to aggregate results and assign each place a score. Casting the problem in the multiple classifier systems framework, we compare several candidate combiners with respect to their performance in the visual localization task. A deeper insight into the potential of the sum and product combiners is provided by testing two extensions of these algebraic rules: threshold and weighted modifications. In addition, a voting method, previously used in robot visual localization, is assessed. All combiners are tested on a visual localization task, carried out on a public dataset. It is experimentally demonstrated that the sum rule extensions globally achieve the best performance. The voting method, whilst competitive to the algebraic rules in their standard form, is shown to be outperformed by both their modified versions.

Robot visual localization through local feature fusion: an evaluation of multiple classifiers combination approaches

In the last decade, local image features have been widely used in robot visual localization. In order to assess image similarity, a strategy exploiting these features compares raw descriptors extracted from the current image with those in the models of places. This paper addresses the ensuing step in this process, where a combining function must be used to aggregate results and assign each place a score. Casting the problem in the multiple classifier systems framework, in this paper we compare several candidate combiners with respect to their performance in the visual localization task. For this evaluation, we selected the most popular methods in the class of non-trained combiners, namely the sum rule and product rule. A deeper insight into the potential of these combiners is provided through a discriminativity analysis involving the algebraic rules and two extensions of these methods: the threshold, as well as the weighted modifications. In addition, a voting method, previously used in robot visual localization, is assessed. Furthermore, we address the process of constructing a model of the environment by describing how the model granularity impacts upon performance. All combiners are tested on a visual localization task, carried out on a public dataset. It is experimentally demonstrated that the sum rule extensions globally achieve the best performance, confirming the general agreement on the robustness of this rule in other classification problems. The voting method, whilst competitive with the product rule in its standard form, is shown to be outperformed by its modified versions.

On the difficulty of feature-based attentional modulations in visual object recognition: A modeling study.

Numerous psychophysical experiments have shown an important role for attentional modulations in vision. Behaviorally, allocation of attention can improve performance in object detection and recognition tasks. At the neural level, attention increases firing rates of neurons in visual cortex whose preferred stimulus is currently attended to. However, it is not yet known how these two phenomena are linked, i.e., how the visual system could be "tuned" in a task-dependent fashion to improve task performance. To answer this question, we performed simulations with the HMAX model of object recognition in cortex [45]. We modulated firing rates of model neurons in accordance with experimental results about effects of feature-based attention on single neurons and measured changes in the model's performance in a variety of object recognition tasks. It turned out that recognition performance could only be improved under very limited circumstances and that attentional influences on the process of object recognition per se tend to display a lack of specificity or raise false alarm rates. These observations lead us to postulate a new role for the observed attention-related neural response modulations.

Object-related activity revealed by functional magnetic resonance imaging in human occipital cortex.

The stages of integration leading from local feature analysis to object recognition were explored in human visual cortex by using the technique of functional magnetic resonance imaging. Here we report evidence for object-related activation. Such activation was located at the lateral-posterior aspect of the occipital lobe, just abutting the posterior aspect of the motion-sensitive area MT/V5, in a region termed the lateral occipital complex (LO). LO showed preferential activation to images of objects, compared to a wide range of texture patterns. This activation was not caused by a global difference in the Fourier spatial frequency content of objects versus texture images, since object images produced enhanced LO activation compared to textures matched in power spectra but randomized in phase. The preferential activation to objects also could not be explained by different patterns of eye movements: similar levels of activation were observed when subjects fixated on the objects and when they scanned the objects with their eyes. Additional manipulations such as spatial frequency filtering and a 4-fold change in visual size did not affect LO activation. These results suggest that the enhanced responses to objects were not a manifestation of low-level visual processing. A striking demonstration that activity in LO is uniquely correlated to object detectability was produced by the "Lincoln" illusion, in which blurring of objects digitized into large blocks paradoxically increases their recognizability. Such blurring led to significant enhancement of LO activation. Despite the preferential activation to objects, LO did not seem to be involved in the final, "semantic," stages of the recognition process. Thus, objects varying widely in their recognizability (e.g., famous faces, common objects, and unfamiliar three-dimensional abstract sculptures) activated it to a similar degree. These results are thus evidence for an intermediate link in the chain of processing stages leading to object recognition in human visual cortex.

Semantic feature-based visual attention model for pedestrian detection

Objective
Pedestrian detection under video surveillance systems has always been a hot topic in computer vision research. These systems are widely used in train stations, airports, large commercial plazas, and other public places. However, pedestrian detection remains difficult because of complex backgrounds. Given its development in recent years, the visual attention mechanism has attracted increasing attention in object detection and tracking research, and previous studies have achieved substantial progress and breakthroughs. We propose a novel pedestrian detection method based on the semantic features under the visual attention mechanism.
Method
The proposed semantic feature-based visual attention model is a spatial-temporal model that consists of two parts: the static visual attention model and the motion visual attention model. The static visual attention model in the spatial domain is constructed by combining bottom-up with top-down attention guidance. Based on the characteristics of pedestrians, the bottom-up visual attention model of Itti is improved by intensifying the orientation vectors of elementary visual features to make the visual saliency map suitable for pedestrian detection. In terms of pedestrian attributes, skin color is selected as a semantic feature for pedestrian detection. The regional and Gaussian models are adopted to construct the skin color model. Skin feature-based visual attention guidance is then proposed to complete the top-down process. The bottom-up and top-down visual attentions are linearly combined using the proper weights obtained from experiments to construct the static visual attention model in the spatial domain. The spatial-temporal visual attention model is then constructed via the motion features in the temporal domain. Based on the static visual attention model in the spatial domain, the frame difference method is combined with optical flowing to detect motion vectors. Filtering is applied to process the field of motion vectors. The saliency of motion vectors can be evaluated via motion entropy to make the selected motion feature more suitable for the spatial-temporal visual attention model.
Result
Standard datasets and practical videos are selected for the experiments. The experiments are performed on a MATLAB R2012a platform. The experimental results show that our spatial-temporal visual attention model demonstrates favorable robustness under various scenes, including indoor train station surveillance videos and outdoor scenes with swaying leaves. Our proposed model outperforms the visual attention model of Itti, the graph-based visual saliency model, the phase spectrum of quaternion Fourier transform model, and the motion channel model of Liu in terms of pedestrian detection. The proposed model achieves a 93% accuracy rate on the test video.
Conclusion
This paper proposes a novel pedestrian method based on the visual attention mechanism. A spatial-temporal visual attention model that uses low-level and semantic features is proposed to calculate the saliency map. Based on this model, the pedestrian targets can be detected through focus of attention shifts. The experimental results verify the effectiveness of the proposed attention model for detecting pedestrians.

The role of auditory cortices in the retrieval of single-trial auditory-visual object memories.

Single-trial encounters with multisensory stimuli affect both memory performance and early-latency brain responses to visual stimuli. Whether and how auditory cortices support memory processes based on single-trial multisensory learning is unknown and may differ qualitatively and quantitatively from comparable processes within visual cortices due to purported differences in memory capacities across the senses. We recorded event-related potentials (ERPs) as healthy adults (n = 18) performed a continuous recognition task in the auditory modality, discriminating initial (new) from repeated (old) sounds of environmental objects. Initial presentations were either unisensory or multisensory; the latter entailed synchronous presentation of a semantically congruent or a meaningless image. Repeated presentations were exclusively auditory, thus differing only according to the context in which the sound was initially encountered. Discrimination abilities (indexed by d') were increased for repeated sounds that were initially encountered with a semantically congruent image versus sounds initially encountered with either a meaningless or no image. Analyses of ERPs within an electrical neuroimaging framework revealed that early stages of auditory processing of repeated sounds were affected by prior single-trial multisensory contexts. These effects followed from significantly reduced activity within a distributed network, including the right superior temporal cortex, suggesting an inverse relationship between brain activity and behavioural outcome on this task. The present findings demonstrate how auditory cortices contribute to long-term effects of multisensory experiences on auditory object discrimination. We propose a new framework for the efficacy of multisensory processes to impact both current multisensory stimulus processing and unisensory discrimination abilities later in time.

Single-trial multisensory memories affect later auditory and visual object discrimination.

Multisensory memory traces established via single-trial exposures can impact subsequent visual object recognition. This impact appears to depend on the meaningfulness of the initial multisensory pairing, implying that multisensory exposures establish distinct object representations that are accessible during later unisensory processing. Multisensory contexts may be particularly effective in influencing auditory discrimination, given the purportedly inferior recognition memory in this sensory modality. The possibility of this generalization and the equivalence of effects when memory discrimination was being performed in the visual vs. auditory modality were at the focus of this study. First, we demonstrate that visual object discrimination is affected by the context of prior multisensory encounters, replicating and extending previous findings by controlling for the probability of multisensory contexts during initial as well as repeated object presentations. Second, we provide the first evidence that single-trial multisensory memories impact subsequent auditory object discrimination. Auditory object discrimination was enhanced when initial presentations entailed semantically congruent multisensory pairs and was impaired after semantically incongruent multisensory encounters, compared to sounds that had been encountered only in a unisensory manner. Third, the impact of single-trial multisensory memories upon unisensory object discrimination was greater when the task was performed in the auditory vs. visual modality. Fourth, there was no evidence for correlation between effects of past multisensory experiences on visual and auditory processing, suggestive of largely independent object processing mechanisms between modalities. We discuss these findings in terms of the conceptual short term memory (CSTM) model and predictive coding. Our results suggest differential recruitment and modulation of conceptual memory networks according to the sensory task at hand.

The Use of Grouping in Visual Object Recognition.

The report describes a recognition system called GROPER, which performs grouping by using distance and relative orientation constraints that estimate the likelihood of different edges in an image coming from the same object. The thesis presents both a theoretical analysis of the grouping problem and a practical implementation of a grouping system. GROPER also uses an indexing module to allow it to make use of knowledge of different objects, any of which might appear in an image. We test GROPER by comparing it to a similar recognition system that does not use grouping.

A Detailed Look at Scale and Translation Invariance in a Hierarchical Neural Model of Visual Object Recognition

The HMAX model has recently been proposed by Riesenhuber & Poggio as a hierarchical model of position- and size-invariant object recognition in visual cortex. It has also turned out to model successfully a number of other properties of the ventral visual stream (the visual pathway thought to be crucial for object recognition in cortex), and particularly of (view-tuned) neurons in macaque inferotemporal cortex, the brain area at the top of the ventral stream. The original modeling study only used ``paperclip'' stimuli, as in the corresponding physiology experiment, and did not explore systematically how model units' invariance properties depended on model parameters. In this study, we aimed at a deeper understanding of the inner workings of HMAX and its performance for various parameter settings and ``natural'' stimulus classes. We examined HMAX responses for different stimulus sizes and positions systematically and found a dependence of model units' responses on stimulus position for which a quantitative description is offered. Interestingly, we find that scale invariance properties of hierarchical neural models are not independent of stimulus class, as opposed to translation invariance, even though both are affine transformations within the image plane.

Geometric Aspects of Visual Object Recognition

This thesis presents there important results in visual object recognition based on shape. (1) A new algorithm (RAST; Recognition by Adaptive Sudivisions of Tranformation space) is presented that has lower average-case complexity than any known recognition algorithm. (2) It is shown, both theoretically and empirically, that representing 3D objects as collections of 2D views (the "View-Based Approximation") is feasible and affects the reliability of 3D recognition systems no more than other commonly made approximations. (3) The problem of recognition in cluttered scenes is considered from a Bayesian perspective; the commonly-used "bounded-error errorsmeasure" is demonstrated to correspond to an independence assumption. It is shown that by modeling the statistical properties of real-scenes better, objects can be recognized more reliably.

Visual Object Recognition Using Deformable Models of Vehicles

This workshop paper reports recent developments to a vision system for traffic interpretation which relies extensively on the use of geometrical and scene context. Firstly, a new approach to pose refinement is reported, based on forces derived from prominent image derivatives found close to an initial hypothesis. Secondly, a parameterised vehicle model is reported, able to represent different vehicle classes. This general vehicle model has been fitted to sample data, and subjected to a Principal Component Analysis to create a deformable model of common car types having 6 parameters. We show that the new pose recovery technique is also able to operate on the PCA model, to allow the structure of an initial vehicle hypothesis to be adapted to fit the prevailing context. We report initial experiments with the model, which demonstrate significant improvements to pose recovery.

Neural mechanisms of visual object priming: evidence for perceptual and semantic distinctions in fusiform cortex

Previous functional imaging studies have shown that facilitated processing of a visual object on repeated, relative to initial, presentation (i.e., repetition priming) is associated with reductions in neural activity in multiple regions, including fusiforin/lateral occipital cortex. Moreover, activity reductions have been found, at diminished levels, when a different exemplar of an object is presented on repetition. In one previous study, the magnitude of diminished priming across exemplars was greater in the right relative to the left fusiform, suggesting greater exemplar specificity in the right. Another previous study, however, observed fusiform lateralization modulated by object viewpoint, but not object exemplar. The present fMRI study sought to determine whether the result of differential fusiform responses for perceptually different exemplars could be replicated. Furthermore, the role of the left fusiform cortex in object recognition was investigated via the inclusion of a lexical/semantic manipulation. Right fusiform cortex showed a significantly greater effect of exemplar change than left fusiform, replicating the previous result of exemplar-specific fusiform lateralization. Right fusiform and lateral occipital cortex were not differentially engaged by the lexical/semantic manipulation, suggesting that their role in visual object recognition is predominantly in the. C visual discrimination of specific objects. Activation in left fusiform cortex, but not left lateral occipital cortex, was modulated by both exemplar change and lexical/semantic manipulation, with further analysis suggesting a posterior-to-anterior progression between regions involved in processing visuoperceptual and lexical/semantic information about objects. The results are consistent with the view that the right fusiform plays a greater role in processing specific visual form information about objects, whereas the left fusiform is also involved in lexical/semantic processing. (C) 2003 Elsevier Science (USA). All rights reserved.