Publish in OALib Journal

ISSN: 2333-9721

APC: Only $99


Any time

2019 ( 37 )

2018 ( 411 )

2017 ( 356 )

2016 ( 313 )

Custom range...

Search Results: 1 - 10 of 23441 matches for " Jifeng He "
All listed articles are free for downloading (OA Articles)
Page 1 /23441
Display every page Item
Instance-aware Semantic Segmentation via Multi-task Network Cascades
Jifeng Dai,Kaiming He,Jian Sun
Computer Science , 2015,
Abstract: Semantic segmentation research has recently witnessed rapid progress, but many leading methods are unable to identify object instances. In this paper, we present Multi-task Network Cascades for instance-aware semantic segmentation. Our model consists of three networks, respectively differentiating instances, estimating masks, and categorizing objects. These networks form a cascaded structure, and are designed to share their convolutional features. We develop an algorithm for the nontrivial end-to-end training of this causal, cascaded structure. Our solution is a clean, single-step training framework and can be generalized to cascades that have more stages. We demonstrate state-of-the-art instance-aware semantic segmentation accuracy on PASCAL VOC. Meanwhile, our method takes only 360ms testing an image using VGG-16, which is two orders of magnitude faster than previous systems for this challenging problem. As a by product, our method also achieves compelling object detection results which surpass the competitive Fast/Faster R-CNN systems. The method described in this paper is the foundation of our submissions to the MS COCO 2015 segmentation competition, where we won the 1st place.
Convolutional Feature Masking for Joint Object and Stuff Segmentation
Jifeng Dai,Kaiming He,Jian Sun
Computer Science , 2014,
Abstract: The topic of semantic segmentation has witnessed considerable progress due to the powerful features learned by convolutional neural networks (CNNs). The current leading approaches for semantic segmentation exploit shape information by extracting CNN features from masked image regions. This strategy introduces artificial boundaries on the images and may impact the quality of the extracted features. Besides, the operations on the raw image domain require to compute thousands of networks on a single image, which is time-consuming. In this paper, we propose to exploit shape information via masking convolutional features. The proposal segments (e.g., super-pixels) are treated as masks on the convolutional feature maps. The CNN features of segments are directly masked out from these maps and used to train classifiers for recognition. We further propose a joint method to handle objects and "stuff" (e.g., grass, sky, water) in the same framework. State-of-the-art results are demonstrated on benchmarks of PASCAL VOC and new PASCAL-CONTEXT, with a compelling computational speed.
BoxSup: Exploiting Bounding Boxes to Supervise Convolutional Networks for Semantic Segmentation
Jifeng Dai,Kaiming He,Jian Sun
Computer Science , 2015,
Abstract: Recent leading approaches to semantic segmentation rely on deep convolutional networks trained with human-annotated, pixel-level segmentation masks. Such pixel-accurate supervision demands expensive labeling effort and limits the performance of deep networks that usually benefit from more training data. In this paper, we propose a method that achieves competitive accuracy but only requires easily obtained bounding box annotations. The basic idea is to iterate between automatically generating region proposals and training convolutional networks. These two steps gradually recover segmentation masks for improving the networks, and vise versa. Our method, called BoxSup, produces competitive results supervised by boxes only, on par with strong baselines fully supervised by masks under the same setting. By leveraging a large amount of bounding boxes, BoxSup further unleashes the power of deep convolutional networks and yields state-of-the-art results on PASCAL VOC 2012 and PASCAL-CONTEXT.
MDM: A Mode Diagram Modeling Framework
Zheng Wang,Geguang Pu,Jianwen Li,Jifeng He
Electronic Proceedings in Theoretical Computer Science , 2012, DOI: 10.4204/eptcs.105.10
Abstract: Periodic control systems used in spacecrafts and automotives are usually period-driven and can be decomposed into different modes with each mode representing a system state observed from outside. Such systems may also involve intensive computing in their modes. Despite the fact that such control systems are widely used in the above-mentioned safety-critical embedded domains, there is lack of domain-specific formal modelling languages for such systems in the relevant industry. To address this problem, we propose a formal visual modeling framework called mode diagram as a concise and precise way to specify and analyze such systems. To capture the temporal properties of periodic control systems, we provide, along with mode diagram, a property specification language based on interval logic for the description of concrete temporal requirements the engineers are concerned with. The statistical model checking technique can then be used to verify the mode diagram models against desired properties. To demonstrate the viability of our approach, we have applied our modelling framework to some real life case studies from industry and helped detect two design defects for some spacecraft control systems.
A Velocity Measurement Method Based on Scaling Parameter Estimation of a Chaotic System
Lidong Liu, Jifeng Hu, Zishu He, Chunlin Han, Huiyong Li, Jun Li
Metrology and Measurement Systems , 2011, DOI: 10.2478/v10178-011-0009-1
Abstract: In this paper, we propose a new method of measuring the target velocity by estimating the scaling parameter of a chaos-generating system. First, we derive the relation between the target velocity and the scaling parameter of the chaos-generating system. Then a new method for scaling parameter estimation of the chaotic system is proposed by exploiting the chaotic synchronization property. Finally, numerical simulations show the effectiveness of the proposed method in target velocity measurement.
Polsat: A Portfolio LTL Satisfiability Solver
Jianwen Li,Geguang Pu,Lijun Zhang,Yinbo Yao,Moshe Y. Vardi,Jifeng he
Computer Science , 2013,
Abstract: In this paper we present a portfolio LTL-satisfiability solver, called Polsat. To achieve fast satisfiability checking for LTL formulas, the tool integrates four representative LTL solvers: pltl, TRP++, NuSMV, and Aalta. The idea of Polsat is to run the component solvers in parallel to get best overall performance; once one of the solvers terminates, it stops all other solvers. Remarkably, the Polsat solver utilizes the power of modern multi-core compute clusters. The empirical experiments show that Polsat takes advantages of it. Further, Polsat is also a testing plat- form for all LTL solvers.
LTLf satisfiability checking
Jianwen Li,Lijun Zhang,Geguang Pu,Moshe Y. Vardi,Jifeng He
Computer Science , 2014,
Abstract: We consider here Linear Temporal Logic (LTL) formulas interpreted over \emph{finite} traces. We denote this logic by LTLf. The existing approach for LTLf satisfiability checking is based on a reduction to standard LTL satisfiability checking. We describe here a novel direct approach to LTLf satisfiability checking, where we take advantage of the difference in the semantics between LTL and LTLf. While LTL satisfiability checking requires finding a \emph{fair cycle} in an appropriate transition system, here we need to search only for a finite trace. This enables us to introduce specialized heuristics, where we also exploit recent progress in Boolean SAT solving. We have implemented our approach in a prototype tool and experiments show that our approach outperforms existing approaches.
On the Relationship between LTL Normal Forms and Buechi Automata
Jianwen Li,Geguang Pu,Lijun Zhang,Zheng Wang,Jifeng He,Kim G. Larsen
Computer Science , 2012,
Abstract: In this paper, we consider the problem of translating LTL formulas to Buechi automata. We first translate the given LTL formula into a special disjuctive-normal form (DNF). The formula will be part of the state, and its DNF normal form specifies the atomic properties that should hold immediately (labels of the transitions) and the formula that should hold afterwards (the corresponding successor state). Surprisingly, if the given formula is Until-free or Release-free, the Buechi automaton can be obtained directly in this manner. For a general formula, the construction is slightly involved: an additional component will be needed for each formula that helps us to identify the set of accepting states. Notably, our construction is an on-the-fly construction, and the resulting Buechi automaton has in worst case 2^{2n+1} states where n denotes the number of subformulas. Moreover, it has a better bound 2^{n+1} when the formula is Until- (or Release-) free.
Fast LTL Satisfiability Checking by SAT Solvers
Jianwen Li,Geguang Pu,Lijun Zhang,Moshe Y. Vardi,Jifeng He
Computer Science , 2014,
Abstract: Satisfiability checking for Linear Temporal Logic (LTL) is a fundamental step in checking for possible errors in LTL assertions. Extant LTL satisfiability checkers use a variety of different search procedures. With the sole exception of LTL satisfiability checking based on bounded model checking, which does not provide a complete decision procedure, LTL satisfiability checkers have not taken advantage of the remarkable progress over the past 20 years in Boolean satisfiability solving. In this paper, we propose a new LTL satisfiability-checking framework that is accelerated using a Boolean SAT solver. Our approach is based on the variant of the \emph{obligation-set method}, which we proposed in earlier work. We describe here heuristics that allow the use of a Boolean SAT solver to analyze the obligations for a given LTL formula. The experimental evaluation indicates that the new approach provides a a significant performance advantage.
Quantum field theory: Finiteness and Effectiveness
Jifeng Yang
Physics , 1998,
Abstract: A new attempt is demonstrated that QFTs can be UV finite if they are viewed as the low energy effective theories of a fundamental underlying theory (that is complete and well-defined in all respects) according to the modern standard point of view. This approach works for any interaction model and space-time dimension. It is much simpler in principle and in technology comparing to any known renormalization program.Unlike the known renormalization methods, the importance of the procedure for defining the ambiguities (corresponding to the choice of the renormalization conditions in the conventional program) is fully appreciated in the new approach. It is shown that the high energy theory(s) or the underlying theory(s) in fact 'stipulates (stipulate)' the low energy and effective ones through these definitions within our approach while all the conventional methods miss this important point. Some simple but important nonperturbative examples are discussed to show the power and plausibility of the new approach. Other related issues (especially the IR problem and the implication of our new approach for the canonical quantization procedure) are briefly touched.
Page 1 /23441
Display every page Item

Copyright © 2008-2017 Open Access Library. All rights reserved.