The proceedings contain 164 papers. The topics discussed include: hallucination space relationship learning to improve very low resolution face recognition;robust local representation for face recognition with single ...
ISBN:
(纸本)9781479961009
The proceedings contain 164 papers. The topics discussed include: hallucination space relationship learning to improve very low resolution face recognition;robust local representation for face recognition with single sample per person;joint space learning for video-based face recognition;DeNet: an explicit distance ensemble model for person re-identification;depth-based person re-identification;structure-driven facade parsing with irregular patterns;unsupervised daily routine modeling from a depth sensor using top-down and bottom-up hierarchies;an extension of PatchMatch stereo for 3D reconstruction from multi-view images;spatial distribution feature for 3D indoor scene labeling;video-level violence rating with rank prediction;color reprint for hypochromatopsia correction;person re-identification using color enhancing feature;and online selection of discriminative features with approximated distribution fields for efficient object tracking.
In this paper, we propose a simple and effective feature learning architecture for image classcation that is based on very basic data processing components: 1) principal component analysis (PCA);2) linear discriminant...
详细信息
ISBN:
(纸本)9781538633540
In this paper, we propose a simple and effective feature learning architecture for image classcation that is based on very basic data processing components: 1) principal component analysis (PCA);2) linear discriminant analysis (LDA);and 3) binary hashing and blockwise histograms. In this architecture, the PCA is employed to reconstruct patches of input images, and the LDA is employed to learn filter banks. This is followed by simple binary hashing and block wise histograms for indexing. This architecture is motivated by LDANet and PCANet [3], thus called the PCA LDA Network (PCA-LDANet). They have some similarities in their topologies. We have tested the PCA-LDANet on two visual datasets for different tasks, including the Facial recognition Technology (FERET) dataset for face recognition;and MNIST dataset for hand-written digit recognition. To explore the properties and essence of these architectures, we just conduct experiments on the one-stage networks. It is enough to explain the issue properly. Experimental results show that the PCA-LDANet-1 outperforms both PCANet-1 and LDANet-1 on both datasets. The experimental results demonstrate the effectiveness and distinctiveness of the PCA-LDANet;and the important role of PCA patch reconstruction in the PCA-LDANet.
The traffic sign recognition system inside the vehicle plays an important role and could guarantee the safety of human life on the road since it feedbacks road information to the driver in time. Benefited from learnin...
详细信息
ISBN:
(纸本)9781538633540
The traffic sign recognition system inside the vehicle plays an important role and could guarantee the safety of human life on the road since it feedbacks road information to the driver in time. Benefited from learning features of the traffic sign, the convolutional neural network (CNN) has been widely used in traffic sign recognition with a high accuracy. However, the different kinds of traffic signs appear to distinctive features. A deep and high complexity neural network with a larger number of parameters is usually required to adapt the distinctive features, while it tends to be time-consuming and can not meet real-time requirement. In this paper, we firstly divide traffic signs into hierarchal structure according to the types of features, and then use a combined CNNs (CCNN) to adapt the hierarchical traffic signs, where the probabilities of superclass and subclass the sign belongs to are calculated using two CNNs with a simple network. Finally, classifying of the sign can be achieved by the weighted output of the two CNNs, and a low complexity sign recognition system could be obtained. Simulation results on the GTSRB database show that the proposed method achieves comparable accuracy and less time-consuming to the state-of-the-art methods.
Aiming at the low prediction accuracy caused by instability of trajectory such as multiple path choices, local abnormal path and flexible step length, an integrated LSTM prediction method based on multi-scale trajecto...
详细信息
ISBN:
(纸本)9781538633540
Aiming at the low prediction accuracy caused by instability of trajectory such as multiple path choices, local abnormal path and flexible step length, an integrated LSTM prediction method based on multi-scale trajectory space (MILSTM) is proposed to predict the coordinate of latitude and longitude. Firstly, the multi-scale fuzzy trajectory space is constructed with the sharing information of similar trajectory to reduce restriction of the road network, and highlight the trajectory intention, meanwhile fuzzy the behavior details in different scales. Then the LSTM models in all scales are integrated by the optimal weight matrix to predict the final coordinates. And the simulation results on trajectory data of Shanghai verified that compared with the classic LSTM model, the expansion of the dataset caused by the fuzzy scale can reduce the prediction error by about 10%, and the multi-scale and integration can effectively suppress the prediction error caused by the trajectory instability, with the increasing instability, the error is reduced by between 10% and 25%.
In this paper we present a new method for writer identification, which extract original Local Binary pattern(LBP) of different radius and Edge descriptors from the edge points of the handwriting. Then, we make combina...
详细信息
We present a novel approach to boost image matching performance by fusing multiple local descriptors in the homography space. Traditional matching methods find correspondences based on a single descriptor and the perf...
详细信息
This paper, for the first time, introduces a multiple-class boosting scheme (MBS) to combine depth motion maps (DMMs) and completed local binary patterns (CLBP) for action recognition. DMMs derive from projecting dept...
详细信息
Observing the world in high speed can give rise to systems for e.g., tracking with super-human capabilities or extremely low latency human-computer interaction. However, the demand on computational power grows with th...
详细信息
A person's routine incorporates the frequent and regular behaviour patterns over a time scale, e.g. daily routine. In this work we present a method for unsupervised discovery of a single person's daily routine...
详细信息
Over-segmentation is often used in text recognition to generate candidate characters. In this paper, we propose a neural network-based over-segmentation method for cropped scene text recognition. On binarized text lin...
详细信息
暂无评论