the 3D restricted Voronoi diagram (RVD), defined as the intersection of the 3D Voronoi diagram of a pointset with a mesh surface, has many applications in geometry processing. there exist several CPU algorithms for co...
详细信息
the upgrades of the Compact Muon Solenoid particle physics experiment at CERN's Large Hadron Collider provide a major challenge for the real-time collision data selection. this paper presents a novel approach to p...
详细信息
the upgrades of the Compact Muon Solenoid particle physics experiment at CERN's Large Hadron Collider provide a major challenge for the real-time collision data selection. this paper presents a novel approach to patternrecognition and charged particle trajectory reconstruction using an all-FPGA solution. the challenges include a large input data rate of about 20 to 40 Tbps, processing a new batch of input data every 25 ns, each consisting of about 10,000 precise position measurements of particles (`stubs'), perform the patternrecognition on these stubs to find the trajectories, and produce the list of parameters describing these trajectories within 4 μs. A proposed solution to this problem is described, in particular, the implementation of the patternrecognition and particle trajectory determination using an all-FPGA system. the results of an end-to-end demonstrator system based on Xilinx Virtex-7 FPGAs that meets timing and performance requirements are presented.
this study explores a unique pose of human walking by proposing centroid based pose ratio for pedestrian action recognition. Our proposal uses width and height of bounding box, and also centroid of human shape to comp...
详细信息
ISBN:
(纸本)9781509008742
this study explores a unique pose of human walking by proposing centroid based pose ratio for pedestrian action recognition. Our proposal uses width and height of bounding box, and also centroid of human shape to compute a pose ratio. then, the proposed model incorporates the pose ratio with pedestrian lateral speed, walking direction and spatial layout of the environment to perform action classification. Lateral speed of pedestrian is determined by using point tracking on the projected point of centroid of detected human on the ground plane. Walking direction of object is estimated by using concatenated consecutive frames. Context of road is used to define the spatial layout. Pedestrian is detected in which location, road or non-road zone. Lane detection is performed for estimating the road boundary as the road zone. the proposed algorithm is evaluated on publicly available datasets. the performance results are presented.
Palindrome recognition is a classic problem in computer science. It is an example of a language that can not be recognized by a deterministic finite automaton and is often brought as an example of a problem whose deci...
详细信息
ISBN:
(纸本)9783319075662;9783319075655
Palindrome recognition is a classic problem in computer science. It is an example of a language that can not be recognized by a deterministic finite automaton and is often brought as an example of a problem whose decision by a single-tape Turing machine requires quadratic time. In this paper we re-visit the palindrome recognition problem. We define a novel fingerprint that allows recognizing palindromes on-line in linear time with high probability. We then use group testing techniques to show that the fingerprint can be adapted to recognizing approximate palindromes on-line, i.e. it can recognize that a string is a palindrome with no more than k mismatches, where k is given. Finally, we show that this fingerprint can be used as a tool for solving other problems on-line. In particular we consider approximate pattern matching by non-overlapping reversals. this is the problem where two strings S and T are given and the question is whether applying a sequence of non-overlapping reversals to S results in string T.
the proceedings contain 28 papers. the topics discussed include: on the efficiency of the hamming C-center string problems;dictionary matching with one gap;approximate on-line palindrome recognition, and applications;...
ISBN:
(纸本)9783319075655
the proceedings contain 28 papers. the topics discussed include: on the efficiency of the hamming C-center string problems;dictionary matching with one gap;approximate on-line palindrome recognition, and applications;computing minimal and maximal suffixes of a substring revisited;compressed subsequence matching and packed tree coloring;reversal distances for strings with few blocks or small alphabets;on combinatorial generation of prefix normal words;permuted scaled matching;the worst case complexity of maximum parsimony;from indexing data structures to de bruijn graphs;randomized and parameterized algorithms for the closest string problem;indexed geometric jumbled pattern matching;an improved query time for succinct dynamic dictionary matching;order-preserving pattern matching with k mismatches;parameterized complexity analysis for the closest string with wildcards problem;and computing palindromic factorizations and palindromic covers on-line.
We propose a novel framework which learns and associates local motion pattern manifolds in streaming videos using generalized regression neural networks (GRNN) to facilitate real time human action recognition. the mot...
详细信息
ISBN:
(纸本)9783319143644;9783319143637
We propose a novel framework which learns and associates local motion pattern manifolds in streaming videos using generalized regression neural networks (GRNN) to facilitate real time human action recognition. the motivation is to determine an individual's action even when the action cycle has not yet been completed. the GRNNs are trained to model the regression function of patterns in latent action space on the input local motion-shape patterns. this manifold learning makes the framework invariant to different sequence length and varying action states. Computation of latent action basis is done using EOF analysis and association of local temporal patterns to an action class at runtime follows a probabilistic formulation. this corresponds to finding the closest estimate the GRNN obtains to the corresponding action basis. Experimental results on two datasets, Kth and the UCF Sports, show accuracy of above 90% obtained from 15 to 25 frames.
Recently, a multilingual Multi Layer Perceptron (MLP) training method was introduced without having to explicitly map the phonetic units of multiple languages to a common set. this paper further investigates this meth...
详细信息
ISBN:
(纸本)9781629934433
Recently, a multilingual Multi Layer Perceptron (MLP) training method was introduced without having to explicitly map the phonetic units of multiple languages to a common set. this paper further investigates this method using bottleneck (BN) tandem connectionist acoustic modeling for four high-resourced languages English, French, german, and Polish Aiming at the improvement of already existing high performing automatic speech recognition (ASR) systems, the multilingual training of the BN-MLP is extended from short-term to hierarchical long-term (multi-resolutional RASTA) feature extraction. Furthermore, deeper structures and context-dependent target labels are also examined. We experimentally demonstrate that a single state-of-the-art BN feature set can be trained for multiple languages, which is superior to the monolingual feature set, and results in significant gains in all the four languages. Studying the scalability of the multilingual BN features, a similar gain is observed in small (50 hours) and in larger scale (300 hours) ASR experiments regardless of the distribution of the data amount between the languages. Using deeper structures, context-dependent targets, and speaker adaptation, the multilingual BN reduces the word error rates by 3-7% relative over the target language BN features and 25-30% over the conventional MFCC system.
the proceedings contain 50 papers. the topics discussed include: as time goes by-anytime semantic segmentation with iterative context forests;interactive labeling of image segmentation hierarchies;hierarchy of localiz...
ISBN:
(纸本)9783642327162
the proceedings contain 50 papers. the topics discussed include: as time goes by-anytime semantic segmentation with iterative context forests;interactive labeling of image segmentation hierarchies;hierarchy of localized random forests for video annotation;curvature prior for MRF-based segmentation and shape inpainting;anisotropic range image integration;modeling of sparsely sampled tubular surfaces using coupled curves;shape (self-)similarity and dissimilarity rating for segmentation and matching;dense 3D reconstruction with a hand-held camera;OUR-CVFH - oriented, unique and repeatable clustered viewpoint feature histogram for object recognition and 6DOF pose estimation;3D object recognition and pose estimation for multiple objects using multi-prioritized RANSAC and model updating;classification with global, local and shared features;and eye localization using the discriminative generalized Hough transform.
In this paper, we present a real-time continuous gesture recognition system for conducting a virtual concert. Our systems allow the user control over beat, by conducting four different beat-pattern gestures;tempo, by ...
详细信息
ISBN:
(纸本)9781450315821
In this paper, we present a real-time continuous gesture recognition system for conducting a virtual concert. Our systems allow the user control over beat, by conducting four different beat-pattern gestures;tempo, by making faster or slower gestures;volume, by making larger or smaller gestures;and instrument emphasis, by directing the gestures towards specific areas of the orchestra on a large display. A recognition accuracy of up to 95% could be achieved for the conducting gestures (beat, tempo, and volume).
We study the task of interactive semantic labeling of a segmentation hierarchy. To this end we propose a framework interleaving two components: an automatic labeling step, based on a Conditional Random Field whose dep...
详细信息
暂无评论