For many computer vision and machine learning problems, large training sets are key for good performance. However, the most computationally expensive part of many computer vision and machine learning ...algorithms consists of finding nearest neighbor matches to high dimensional vectors that represent the training data. We propose new algorithms for approximate nearest neighbor matching and evaluate and compare them with previous algorithms. For matching high dimensional features, we find two algorithms to be the most efficient: the randomized k-d forest and a new algorithm proposed in this paper, the priority search k-means tree. We also propose a new algorithm for matching binary features by searching multiple hierarchical clustering trees and show it outperforms methods typically used in the literature. We show that the optimal nearest neighbor algorithm and its parameters depend on the data set characteristics and describe an automated configuration procedure for finding the best algorithm to search a particular data set. In order to scale to very large data sets that would otherwise not fit in the memory of a single machine, we propose a distributed nearest neighbor matching framework that can be used with any of the algorithms described in the paper. All this research has been released as an open source library called fast library for approximate nearest neighbors (FLANN), which has been incorporated into OpenCV and is now one of the most popular libraries for nearest neighbor matching.
This paper presents a method for extracting distinctive invariant features from images that can be used to perform reliable matching between different views of an object or scene. The features are ...invariant to image scale and rotation, and are shown to provide robust matching across a substantial range of affine distortion, change in 3D viewpoint, addition of noise, and change in illumination. The features are highly distinctive, in the sense that a single feature can be correctly matched with high probability against a large database of features from many images. This paper also describes an approach to using these features for object recognition. The recognition proceeds by matching individual features to a database of features from known objects using a fast nearest-neighbor algorithm, followed by a Hough transform to identify clusters belonging to a single object, and finally performing verification through least-squares solution for consistent pose parameters. This approach to recognition can robustly identify objects among clutter and occlusion while achieving near real-time performance.PUBLICATION ABSTRACT
This paper concerns the problem of fully automated panoramic image stitching. Though the 1D problem (single axis of rotation) is well studied, 2D or multi-row stitching is more difficult. Previous ...approaches have used human input or restrictions on the image sequence in order to establish matching images. In this work, we formulate stitching as a multi-image matching problem, and use invariant local features to find matches between all of the images. Because of this our method is insensitive to the ordering, orientation, scale and illumination of the input images. It is also insensitive to noise images that are not part of a panorama, and can recognise multiple panoramas in an unordered image dataset. In addition to providing more detail, this paper extends our previous work in the area (Brown and Lowe, 2003) by introducing gain compensation and automatic straightening steps.
We present an unsupervised learning framework for the task of monocular depth and camera motion estimation from unstructured video sequences. In common with recent work 10, 14, 16, we use an ...end-to-end learning approach with view synthesis as the supervisory signal. In contrast to the previous work, our method is completely unsupervised, requiring only monocular video sequences for training. Our method uses single-view depth and multiview pose networks, with a loss based on warping nearby views to the target using the computed depth and pose. The networks are thus coupled by the loss during training, but can be applied independently at test time. Empirical evaluation on the KITTI dataset demonstrates the effectiveness of our approach: 1) monocular depth performs comparably with supervised methods that use either ground-truth pose or depth for training, and 2) pose estimation performs favorably compared to established SLAM systems under comparable input settings.
We investigate the role of sparsity and localized features in a biologically-inspired model of visual object classification. As in the model of Serre, Wolf, and Poggio, we first apply Gabor filters ...at all positions and scales; feature complexity and position/scale invariance are then built up by alternating template matching and max pooling operations. We refine the approach in several biologically plausible ways. Sparsity is increased by constraining the number of feature inputs, lateral inhibition, and feature selection. We also demonstrate the value of retaining some position and scale information above the intermediate feature level. Our final model is competitive with current computer vision algorithms on several standard datasets, including the Caltech 101 object categories and the UIUC car localization task. The results further the case for biologically-motivated approaches to object classification.
The aim of this study was to improve the neuropathologic recognition and provide criteria for the pathological diagnosis in the neurodegenerative diseases grouped as frontotemporal lobar degeneration ...(FTLD); revised criteria are proposed. Recent advances in molecular genetics, biochemistry, and neuropathology of FTLD prompted the Midwest Consortium for Frontotemporal Lobar Degeneration and experts at other centers to review and revise the existing neuropathologic diagnostic criteria for FTLD. The proposed criteria for FTLD are based on existing criteria, which include the tauopathies FTLD with Pick bodies, corticobasal degeneration, progressive supranuclear palsy, sporadic multiple system tauopathy with dementia, argyrophilic grain disease, neurofibrillary tangle dementia, and FTD with microtubule-associated tau (MAPT) gene mutation, also called FTD with parkinsonism linked to chromosome 17 (FTDP-17). The proposed criteria take into account new disease entities and include the novel molecular pathology, TDP-43 proteinopathy, now recognized to be the most frequent histological finding in FTLD. TDP-43 is a major component of the pathologic inclusions of most sporadic and familial cases of FTLD with ubiquitin-positive, tau-negative inclusions (FTLD-U) with or without motor neuron disease (MND). Molecular genetic studies of familial cases of FTLD-U have shown that mutations in the progranulin (PGRN) gene are a major genetic cause of FTLD-U. Mutations in valosin-containing protein (VCP) gene are present in rare familial forms of FTD, and some families with FTD and/or MND have been linked to chromosome 9p, and both are types of FTLD-U. Thus, familial TDP-43 proteinopathy is associated with defects in multiple genes, and molecular genetics is required in these cases to correctly identify the causative gene defect. In addition to genetic heterogeneity amongst the TDP-43 proteinopathies, there is also neuropathologic heterogeneity and there is a close relationship between genotype and FTLD-U subtype. In addition to these recent significant advances in the neuropathology of FTLD-U, novel FTLD entities have been further characterized, including neuronal intermediate filament inclusion disease. The proposed criteria incorporate up-to-date neuropathology of FTLD in the light of recent immunohistochemical, biochemical, and genetic advances. These criteria will be of value to the practicing neuropathologist and provide a foundation for clinical, clinico-pathologic, mechanistic studies and in vivo models of pathogenesis of FTLD.
We have previously developed a mobile robot system which uses scale-invariant visual landmarks to localize and simultaneously build three-dimensional (3-D) maps of unmodified environments. In this ...paper, we examine global localization, where the robot localizes itself globally, without any prior location estimate. This is achieved by matching distinctive visual landmarks in the current frame to a database map. A Hough transform approach and a RANSAC approach for global localization are compared, showing that RANSAC is much more efficient for matching specific features, but much worse for matching nonspecific features. Moreover, robust global localization can be achieved by matching a small submap of the local region built from multiple frames. This submap alignment algorithm for global localization can be applied to map building, which can be regarded as alignment of multiple 3-D submaps. A global minimization procedure is carried out using the loop closure constraint to avoid the effects of slippage and drift accumulation. Landmark uncertainty is taken into account in the submap alignment and the global minimization process. Experiments show that global localization can be achieved accurately using the scale-invariant landmarks. Our approach of pairwise submap alignment with backward correction in a consistent manner produces a better global 3-D map.
The broadly conserved signaling nucleotide cyclic di-adenosine monophosphate (c-di-AMP) is essential for viability in most bacteria where it has been studied. However, characterization of the ...cellular functions and metabolism of c-di-AMP has largely been confined to the class Bacilli, limiting our functional understanding of the molecule among diverse phyla. We identified the cyclase responsible for c-di-AMP synthesis and characterized the molecule's role in survival of darkness in the model photosynthetic cyanobacterium Synechococcus elongatus PCC 7942. In addition to the use of traditional genetic, biochemical, and proteomic approaches, we developed a high-throughput genetic interaction screen (IRB-Seq) to determine pathways where the signaling nucleotide is active. We found that in S. elongatus c-di-AMP is produced by an enzyme of the diadenylate cyclase family, CdaA, which was previously unexplored experimentally. A cdaA-null mutant experiences increased oxidative stress and death during the nighttime portion of day-night cycles, in which potassium transport is implicated. These findings suggest that c-di-AMP is biologically active in cyanobacteria and has non-canonical roles in the phylum including oxidative stress management and day-night survival. The pipeline and analysis tools for IRB-Seq developed for this study constitute a quantitative high-throughput approach for studying genetic interactions.
Celotno besedilo
Dostopno za:
DOBA, IZUM, KILJ, NUK, PILJ, PNG, SAZU, SIK, UILJ, UKNU, UL, UM, UPUK
As part of an ongoing search for genes associated with type 1 diabetes (T1D), a common autoimmune disease, we tested the biological candidate gene
IL2RA (
CD25), which encodes a subunit (IL-2Rα) of ...the high-affinity interleukin-2 (IL-2) receptor complex. We employed a tag single-nucleotide polymorphism (tag SNP) approach in large T1D sample collections consisting of 7,457 cases and controls and 725 multiplex families. Tag SNPs were analyzed using a multilocus test to provide a regional test for association. We found strong statistical evidence in the case-control collection (
P
=
6.5×
10
−8) for a T1D locus in the
CD25 region of chromosome 10p15 and replicated the association in the family collection (
P
=
7.3×
10
−3; combined
P
=
1.3×
10
−10). These results illustrate the utility of tag SNPs in a chromosome-regional test of disease association and justify future fine mapping of the causal variant in the region.
Genome-wide association studies are now identifying disease-associated chromosome regions. However, even after convincing replication, the localization of the causal variant(s) requires comprehensive ...resequencing, extensive genotyping and statistical analyses in large sample sets leading to targeted functional studies. Here, we have localized the type 1 diabetes (T1D) association in the interleukin 2 receptor alpha (IL2RA) gene region to two independent groups of SNPs, spanning overlapping regions of 14 and 40 kb, encompassing IL2RA intron 1 and the 5' regions of IL2RA and RBM17 (odds ratio = 2.04, 95% confidence interval = 1.70-2.45; P = 1.92 x 10(-28); control frequency = 0.635). Furthermore, we have associated IL2RA T1D susceptibility genotypes with lower circulating levels of the biomarker, soluble IL-2RA (P = 6.28 x 10(-28)), suggesting that an inherited lower immune responsiveness predisposes to T1D.
Celotno besedilo
Dostopno za:
DOBA, IJS, IZUM, KILJ, NUK, PILJ, PNG, SAZU, UILJ, UKNU, UL, UM, UPUK