Multi-scale label interaction to improve object detection and segmentation

The project aims at advancing the state of the art in the use of context information for classification of image and video data. The use of context in the classification of images has been showed of great importance to improve the performance of actual object recognition systems. In our project we proposed the concept of Multi-scale Feature Labels as a general and compact method to exploit the local and global context. The feature extraction from the discriminative probability or classification confidence label field is of great novelty. Moreover the use of a multi-scale representation of the feature labels lead to a compact and efficient description of the context. The goal of the project has been also to provide a general-purpose method and prove its suitability in different image/video analysis problem. The two-year project generated 5 journal publications (plus 2 under submission), 10 conference publications (plus 2 under submission) and one patent (plus 1 pending). Of these publications, a relevant number make use of the main result of this project to improve the results in detection and/or segmentation of objects.

Motion detection and motion verbs: language affects low-level visual perception

Recent theories propose that semantic representation and sensorimotor processing have a common substrate via simulation. We tested the prediction that comprehension interacts with perception, using a standard psychophysics methodology.While passively listening to verbs that referred to upward or downward motion, and to control verbs that did not refer to motion, 20 subjects performed a motion-detection task, indicating whether or not they saw motion in visual stimuli containing threshold levels of coherent vertical motion. A signal detection analysis revealed that when verbs were directionally incongruent with the motion signal, perceptual sensitivity was impaired. Word comprehension also affected decision criteria and reaction times, but in different ways. The results are discussed with reference to existing explanations of embodied processing and the potential of psychophysical methods for assessing interactions between language and perception.

Marine mammal species detection and classification

Thesis (Ph.D.)--University of Washington, 2016-04

Multisensorial obstacles detection and classification for ADAS and autonomous driving

A reliable perception of the real world is a key-feature for an autonomous vehicle and the Advanced Driver Assistance Systems (ADAS). Obstacles detection (OD) is one of the main components for the correct reconstruction of the dynamic world. Historical approaches based on stereo vision and other 3D perception technologies (e.g. LIDAR) have been adapted to the ADAS first and autonomous ground vehicles, after, providing excellent results. The obstacles detection is a very broad field and this domain counts a lot of works in the last years. In academic research, it has been clearly established the essential role of these systems to realize active safety systems for accident prevention, reflecting also the innovative systems introduced by industry. These systems need to accurately assess situational criticalities and simultaneously assess awareness of these criticalities by the driver; it requires that the obstacles detection algorithms must be reliable and accurate, providing: a real-time output, a stable and robust representation of the environment and an estimation independent from lighting and weather conditions. Initial systems relied on only one exteroceptive sensor (e.g. radar or laser for ACC and camera for LDW) in addition to proprioceptive sensors such as wheel speed and yaw rate sensors. But, current systems, such as ACC operating at the entire speed range or autonomous braking for collision avoidance, require the use of multiple sensors since individually they can not meet these requirements. It has led the community to move towards the use of a combination of them in order to exploit the benefits of each one. Pedestrians and vehicles detection are ones of the major thrusts in situational criticalities assessment, still remaining an active area of research. ADASs are the most prominent use case of pedestrians and vehicles detection. Vehicles should be equipped with sensing capabilities able to detect and act on objects in dangerous situations, where the driver would not be able to avoid a collision. A full ADAS or autonomous vehicle, with regard to pedestrians and vehicles, would not only include detection but also tracking, orientation, intent analysis, and collision prediction. The system detects obstacles using a probabilistic occupancy grid built from a multi-resolution disparity map. Obstacles classification is based on an AdaBoost SoftCascade trained on Aggregate Channel Features. A final stage of tracking and fusion guarantees stability and robustness to the result.

Detection and Classification of Whale Acoustic Signals

This dissertation focuses on two vital challenges in relation to whale acoustic signals: detection and classification.

In detection, we evaluated the influence of the uncertain ocean environment on the spectrogram-based detector, and derived the likelihood ratio of the proposed Short Time Fourier Transform detector. Experimental results showed that the proposed detector outperforms detectors based on the spectrogram. The proposed detector is more sensitive to environmental changes because it includes phase information.

In classification, our focus is on finding a robust and sparse representation of whale vocalizations. Because whale vocalizations can be modeled as polynomial phase signals, we can represent the whale calls by their polynomial phase coefficients. In this dissertation, we used the Weyl transform to capture chirp rate information, and used a two dimensional feature set to represent whale vocalizations globally. Experimental results showed that our Weyl feature set outperforms chirplet coefficients and MFCC (Mel Frequency Cepstral Coefficients) when applied to our collected data.

Since whale vocalizations can be represented by polynomial phase coefficients, it is plausible that the signals lie on a manifold parameterized by these coefficients. We also studied the intrinsic structure of high dimensional whale data by exploiting its geometry. Experimental results showed that nonlinear mappings such as Laplacian Eigenmap and ISOMAP outperform linear mappings such as PCA and MDS, suggesting that the whale acoustic data is nonlinear.

We also explored deep learning algorithms on whale acoustic data. We built each layer as convolutions with either a PCA filter bank (PCANet) or a DCT filter bank (DCTNet). With the DCT filter bank, each layer has different a time-frequency scale representation, and from this, one can extract different physical information. Experimental results showed that our PCANet and DCTNet achieve high classification rate on the whale vocalization data set. The word error rate of the DCTNet feature is similar to the MFSC in speech recognition tasks, suggesting that the convolutional network is able to reveal acoustic content of speech signals.

Contributions to 3D Data Registration and Representation

Nowadays, new computers generation provides a high performance that enables to build computationally expensive computer vision applications applied to mobile robotics. Building a map of the environment is a common task of a robot and is an essential part to allow the robots to move through these environments. Traditionally, mobile robots used a combination of several sensors from different technologies. Lasers, sonars and contact sensors have been typically used in any mobile robotic architecture, however color cameras are an important sensor due to we want the robots to use the same information that humans to sense and move through the different environments. Color cameras are cheap and flexible but a lot of work need to be done to give robots enough visual understanding of the scenes. Computer vision algorithms are computational complex problems but nowadays robots have access to different and powerful architectures that can be used for mobile robotics purposes. The advent of low-cost RGB-D sensors like Microsoft Kinect which provide 3D colored point clouds at high frame rates made the computer vision even more relevant in the mobile robotics field. The combination of visual and 3D data allows the systems to use both computer vision and 3D processing and therefore to be aware of more details of the surrounding environment. The research described in this thesis was motivated by the need of scene mapping. Being aware of the surrounding environment is a key feature in many mobile robotics applications from simple robotic navigation to complex surveillance applications. In addition, the acquisition of a 3D model of the scenes is useful in many areas as video games scene modeling where well-known places are reconstructed and added to game systems or advertising where once you get the 3D model of one room the system can add furniture pieces using augmented reality techniques. In this thesis we perform an experimental study of the state-of-the-art registration methods to find which one fits better to our scene mapping purposes. Different methods are tested and analyzed on different scene distributions of visual and geometry appearance. In addition, this thesis proposes two methods for 3d data compression and representation of 3D maps. Our 3D representation proposal is based on the use of Growing Neural Gas (GNG) method. This Self-Organizing Maps (SOMs) has been successfully used for clustering, pattern recognition and topology representation of various kind of data. Until now, Self-Organizing Maps have been primarily computed offline and their application in 3D data has mainly focused on free noise models without considering time constraints. Self-organising neural models have the ability to provide a good representation of the input space. In particular, the Growing Neural Gas (GNG) is a suitable model because of its flexibility, rapid adaptation and excellent quality of representation. However, this type of learning is time consuming, specially for high-dimensional input data. Since real applications often work under time constraints, it is necessary to adapt the learning process in order to complete it in a predefined time. This thesis proposes a hardware implementation leveraging the computing power of modern GPUs which takes advantage of a new paradigm coined as General-Purpose Computing on Graphics Processing Units (GPGPU). Our proposed geometrical 3D compression method seeks to reduce the 3D information using plane detection as basic structure to compress the data. This is due to our target environments are man-made and therefore there are a lot of points that belong to a plane surface. Our proposed method is able to get good compression results in those man-made scenarios. The detected and compressed planes can be also used in other applications as surface reconstruction or plane-based registration algorithms. Finally, we have also demonstrated the goodness of the GPU technologies getting a high performance implementation of a CAD/CAM common technique called Virtual Digitizing.

Development of machine learning methods for multi-modal biomarkers detection and integration

In medicine, innovation depends on a better knowledge of the human body mechanism, which represents a complex system of multi-scale constituents. Unraveling the complexity underneath diseases proves to be challenging. A deep understanding of the inner workings comes with dealing with many heterogeneous information. Exploring the molecular status and the organization of genes, proteins, metabolites provides insights on what is driving a disease, from aggressiveness to curability. Molecular constituents, however, are only the building blocks of the human body and cannot currently tell the whole story of diseases. This is why nowadays attention is growing towards the contemporary exploitation of multi-scale information. Holistic methods are then drawing interest to address the problem of integrating heterogeneous data. The heterogeneity may derive from the diversity across data types and from the diversity within diseases. Here, four studies conducted data integration using customly designed workflows that implement novel methods and views to tackle the heterogeneous characterization of diseases. The first study devoted to determine shared gene regulatory signatures for onco-hematology and it showed partial co-regulation across blood-related diseases. The second study focused on Acute Myeloid Leukemia and refined the unsupervised integration of genomic alterations, which turned out to better resemble clinical practice. In the third study, network integration for artherosclerosis demonstrated, as a proof of concept, the impact of network intelligibility when it comes to model heterogeneous data, which showed to accelerate the identification of new potential pharmaceutical targets. Lastly, the fourth study introduced a new method to integrate multiple data types in a unique latent heterogeneous-representation that facilitated the selection of important data types to predict the tumour stage of invasive ductal carcinoma. The results of these four studies laid the groundwork to ease the detection of new biomarkers ultimately beneficial to medical practice and to the ever-growing field of Personalized Medicine.

Aeromonas detection and their toxins from drinking water form reservoirs and drinking fountains

Aeromonads are inhabitants of aquatic ecosystems and are described as being involved in intestinal disturbances and other infections. A total of 200 drinking water samples from domestic and public reservoirs and drinking fountains located in São Paulo (Brazil), were analyzed for the presence of Aeromonas. Samples were concentrated by membrane filtration and enriched in APW. ADA medium was used for Aeromonas isolation and colonies were confirmed by biochemical characterization. Strains isolated were tested for hemolysin and toxin production. Aeromonas was detected in 12 samples (6.0%). Aeromonas strains (96) were isolated and identified as: A. caviae (41.7%), A. hydrophila (15.7%), A.allosacharophila (10.4%), A. schubertii (1.0%) and Aeromonas spp. (31.2%).The results revealed that 70% of A. caviae, 66.7% of A. hydrophila, 80% of A. allosacharophila and 46.6% of Aeromonas spp. were hemolytic. The assay for checking production of toxins showed that 17.5% of A. caviae, 73.3% of A. hydrophila, 60% of A. allosacharophila, 100% of A. schubertii, and 33.3% of Aeromonas spp. were able to produce toxins. The results demonstrated the pathogenic potential of Aeromonas, indicating that the presence of this emerging pathogen in water systems is a public health concern

Innovation concept for measurement gross error detection and identification in power system state estimation

In this study, the innovation approach is used to estimate the measurement total error associated with power system state estimation. This is required because the power system equations are very much correlated with each other and as a consequence part of the measurements errors is masked. For that purpose an index, innovation index (II), which provides the quantity of new information a measurement contains is proposed. A critical measurement is the limit case of a measurement with low II, it has a zero II index and its error is totally masked. In other words, that measurement does not bring any innovation for the gross error test. Using the II of a measurement, the masked gross error by the state estimation is recovered; then the total gross error of that measurement is composed. Instead of the classical normalised measurement residual amplitude, the corresponding normalised composed measurement residual amplitude is used in the gross error detection and identification test, but with m degrees of freedom. The gross error processing turns out to be very simple to implement, requiring only few adaptations to the existing state estimation software. The IEEE-14 bus system is used to validate the proposed gross error detection and identification test.

Jointly multi-user detection and channel estimation with genetic algorithm

This work aims at proposing the use of the evolutionary computation methodology in order to jointly solve the multiuser channel estimation (MuChE) and detection problems at its maximum-likelihood, both related to the direct sequence code division multiple access (DS/CDMA). The effectiveness of the proposed heuristic approach is proven by comparing performance and complexity merit figures with that obtained by traditional methods found in literature. Simulation results considering genetic algorithm (GA) applied to multipath, DS/CDMA and MuChE and multi-user detection (MuD) show that the proposed genetic algorithm multi-user channel estimation (GAMuChE) yields a normalized mean square error estimation (nMSE) inferior to 11%, under slowly varying multipath fading channels, large range of Doppler frequencies and medium system load, it exhibits lower complexity when compared to both maximum likelihood multi-user channel estimation (MLMuChE) and gradient descent method (GrdDsc). A near-optimum multi-user detector (MuD) based on the genetic algorithm (GAMuD), also proposed in this work, provides a significant reduction in the computational complexity when compared to the optimum multi-user detector (OMuD). In addition, the complexity of the GAMuChE and GAMuD algorithms were (jointly) analyzed in terms of number of operations necessary to reach the convergence, and compared to other jointly MuChE and MuD strategies. The joint GAMuChE-GAMuD scheme can be regarded as a promising alternative for implementing third-generation (3G) and fourth-generation (4G) wireless systems in the near future. Copyright (C) 2010 John Wiley & Sons, Ltd.

Molecular detection and differentiation of Brazilian and African isolates of the entomopathogen Neozygites tanajoae (Entomophthorales: Neozygitaceae) with PCR using specific primers

Neozygites tanajoae is an entomopathogenic fungus which has been used for biocontrol of the cassava green mite (Mononychellus tanajoa, CGM) in Africa. Establishment and dispersal of Brazilian isolates which have been introduced into some African countries in recent years to improve CGM control was followed with specific PCR assays. Two primer pairs, NEOSSU_F/NEOSSU_R and 8DDC_F/8DDC_R, were used to differentiate isolates collected from several locations in Brazil and from three countries in Africa, Benin, Ghana and Tanzania. The first primer pair enabled the species-specific detection of Neozygites tanajoae, while the second differentiated the Brazilian isolates from those of other geographical origin. PCR assays were designed for detection of fungal DNA in the matrix of dead infested mites since N. tanajoae is difficult to isolate and culture on selective artificial media. Our results show that all isolates (Brazilian and African) that sporulated on mummified mites were amplified with the first primer pair confirming their Neozygites tanajoae identity. The second pair amplified DNA from all the Brazilian isolates, but did not amplify any DNA samples from the African isolates. None of the two primers showed amplification neither from any of the non-sporulating mite extracts nor from the dead uninfected mites used as negative controls. We confirmed that the two primer pairs tested are suitable for the detection and differential identification of N. tanajoae isolates from Brazil and Africa and that they are useful to monitor the establishment and spread of the Brazilian isolates of N. tanajoae introduced into Benin or into other African countries for improvement of CGM biocontrol.

NEURAL CORRELATES OF SCENT MARKING BEHAVIOR IN C57BL/6J MICE: DETECTION AND RECOGNITION OF A SOCIAL STIMULUS

Mice show urinary scent marking behavior as a form of social communication. Marking to a conspecific stimulus mouse or odor varies with stimulus familiarity, indicating discrimination of novel and familiar animals. This study investigated Fos immunoreactivity in inbred C57BL/6J (C57) males following scent marking behavior in response to detection of a social stimulus, or discrimination between a familiar and an unfamiliar conspecific. In Experiment 1 C57 mice were exposed for four daily trials to an empty chamber; on a test day they were exposed to the same chamber or to a male CD-1 mouse in that chamber. Increased scent marking to the CD-1 mouse was associated with increased Fos-immunoreactive cells in the basolateral amygdala, medial amygdala, and dorsal and ventral premammillary nuclei. In Experiment 2 C57 mice were habituated to a CD-1 male for 4 consecutive days and, on the 5th day, exposed to the same CD-1 male, or to a novel CD-1 male. Mice exposed to a novel CD-1 displayed a significant increase in scent marking compared to their last exposure to the familiar stimulus, indicating discrimination of the novelty of this social stimulus. Marking to the novel stimulus was associated with enhanced activation of several telencephalic, as well as hypothalamic and midbrain, structures in which activation had not been seen in the detection paradigm (Experiment 1). These included medial prefrontal and piriform cortices, and lateral septum; the paraventricular nuclei, ventromedial nuclei, and lateral area of the hypothalamus, and the ventrolateral column of the periaqueductal gray. These data suggest that a circumscribed group of structures largely concerned with olfaction is involved in detection of a conspecific olfactory stimulus, whereas discrimination of a novel vs. a familiar conspecific stimulus engages a wider range of forebrain structures encompassing higher-order processes and potentially providing an interface between cognitions and emotions. (C) 2009 IBRO. Published by Elsevier Ltd. All rights reserved.

Genetic polymorphism in an inflammasome component, cervical mycoplasma detection and female infertility in women undergoing in vitro fertilization

The inflammasome is an inducible cytoplasmic structure that is responsible for production and release of biologically active interleukin-1 (IL-1). A polymorphism in the inflammasome component NALP3 has been associated with decreased IL-1 levels and increased occurrence of vaginal Candida infection. We hypothesized that this polymorphism-induced variation would influence susceptibility to infertility. DNA was obtained from 243 women who were undergoing in vitro fertilization (IVF) and tested for a length polymorphism in intron 2 of the gene coding for NALP3 (gene symbol CIAS1). At the conclusion of testing the findings were analyzed in relation to clinical parameters and IVF outcome. The frequency of the 12 unit repeat allele, associated with maximal inflammasome activity, was 62.3% in cases of female infertility vs. 75.6% in cases where only the male partner had a detectable fertility problem (p = 0.0095). Conversely, the frequency of the 7 unit repeat allele was 28.9% in those with a female fertility problem, 17.0% in women with infertile males and 18.4% in idiopathic infertility (p = 0.0124). Among the women who were cervical culture-positive for mycoplasma the frequency of the 7 unit repeat was 53.7% as opposed to 19.5% in those negative for this infection (p < 0.0001). We conclude that the CIAS1 7 unit repeat polymorphism increases the likelihood of mycoplasma infection-associated female infertility. (C) 2009 Elsevier Ireland Ltd. All rights reserved.

Cancer detection and mammogram volume of radiologists in a population-based screening programme

This study investigates the relationship between the number of screening mammograms read by radiologists and the screening breast cancer detection rate. Cancer detection rates for incident screens (all women aged >= 40 years) were compared by increasing categories of reader volume using Poisson regression. Data from New South Wales (NSW) for a 2 year period (2000-2001) were obtained from the BreastScreen NSW programme. Cancer detection rates increased with the number of mammograms read in the programme, reaching a plateau of approximately 40 per 10,000 after 1375 mammograms per year. No significant differences in cancer detection were evident above 875 mammograms (compared to below 875 mammograms) per year (RR = 0.79, 95% CI 0.63-0.99). (c) 2005 Elsevier Ltd. All rights reserved.