Compact discriminative object representation via weakly supervised learning for real‐time visual tracking

Shen, Wei Feng; Wu, Yuwei; Jia, Yingmin

doi:10.1049/iet-cvi.2016.0287

Cited by 5 publications

(5 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…CCA looks for the first pair of projections w x and w y which can be found in the objective function (9) in order to maximise the correlations among w x T x and w y T y [62,65] max…”

Section: Multi-view Dimension Reductionmentioning

confidence: 99%

See 1 more Smart Citation

Dictionary‐induced least squares framework for multi‐view dimensionality reduction with multi‐manifold embeddings

Abeo

Shen

Gou

et al. 2019

IET Computer Vision

View full text Add to dashboard Cite

This study proposes a novel dimensionality reduction (DR) method for multi-view datasets. The principal component analysis (PCA) idea of minimising least squares reconstruction errors is extended to consider both data distribution and penalty weights called dictionary to recover outliers free global structures from missing and noisy data points. In this way, PCA is viewed as a special instance of the authors' proposed dictionary induced least squares framework (DLS). Furthermore, to appropriately handle multi-view DR, we combine the DLS with multiple manifold embeddings (DLSME). Therefore it can obtain lower projections while maintaining a balance between preserving global structures with DLS and local structures with multi-manifold embeddings. Extensive experiments on object and face recognition datasets verify that the DLS achieves better classification results with lower dimensional projections than PCA. Also, on many multi-view datasets of visual recognition and web image annotation, the DLSME method demonstrates more effectiveness than Graph-Laplacian PCA (gLPCA), robust PCA-optimal mean, canonical correlation analysis (CCA), bilinear models (BLM), neighbourhood preserving embedding, locality preserving projections, and locality sensitive discriminant analysis.

show abstract

“…CCA looks for the first pair of projections w x and w y which can be found in the objective function (9) in order to maximise the correlations among w x T x and w y T y [62,65] max…”

Section: Multi-view Dimension Reductionmentioning

confidence: 99%

“…This presents a situation of dimensionality curse [6,7]. Therefore numerous methods have been proposed for DR [4,8,9], as well as methods of neural networks [10][11][12][13][14][15][16] in recent times for better representation learning.…”

Section: Introductionmentioning

confidence: 99%

Dictionary‐induced least squares framework for multi‐view dimensionality reduction with multi‐manifold embeddings

Abeo

Shen

Gou

et al. 2019

IET Computer Vision

View full text Add to dashboard Cite

show abstract

“…In the upcoming surveillance networks, the embedded content analysis will play its significant role in smart cameras [1]. As a main building blocks of smart surveillance functions video object tracking as well as segmentation get its imperative attention [2][3][4][5][6]. Normally, the automated object detection is the initial task in the system of multi-camera surveillance and subsequently, the Background Modeling (BM) is a common technique for the extraction of predefined information including shape of the objects, geometry and so on for further processing [7][8][9][10].…”

Section: Introductionmentioning

confidence: 99%

On the Real Time Object Detection and Tracking

Barekar¹,

Singh²,

Kamble³

et al. 2022

JOAASR

View full text Add to dashboard Cite

Object detection and tracking is widely used for detecting motions of objects present in images and video.Since last so many decades, numerous real time object detection and tracking methods have been proposed byresearchers. The proposed methods for objects to be tracked till date require some preceding informationassociated with moving objects. In real time object detection and tracking approach segmentation is the initialtask followed by background modeling for the extraction of predefined information including shape of the objects,position in the starting frame, texture, geometry and so on for further processing of the cluster pixels and videosequence of these objects. The object detection and tracking can be applied in the fields like computerized videosurveillance, traffic monitoring, robotic vision, gesture identification, human-computer interaction, militarysurveillance system, vehicle navigation, medical imaging, biomedical image analysis and many more. In thispaper we focus detailed technical review of different methods proposed for detection and tracking of objects. Thecomparison of various techniques of detection and tracking is the purpose of this work.

show abstract

“…Generative approaches search the most similar region in an image to detect the target model. Researchers have developed several outlines such as pixel intensity [2], contours [3, 4], histograms [5, 6], optical flow [7, 8], point [9, 10] and texture [11, 12] to represent the target appearance. Nevertheless, when the target is occluded and under varying scene illumination, the appearance of an object is not represented well.…”

Section: Introductionmentioning

confidence: 99%

Approach to model human appearance based on sparse representation for human tracking in surveillance

Damotharasamy¹

2020

IET image process

View full text Add to dashboard Cite

In human tracking, sparse representation successfully localises the human in a video with minimal reconstruction error using target templates. However, the state‐of‐the‐art approaches use colour and local appearance of a human to discriminate the human from the background regions, and hence fail when the human is occluded and appears in the varying illumination environment. In this study, a robust tracking algorithm is proposed that utilises gradient orientation and fine and coarse sparse representation of the target template. Sparse representation‐based human appearance model utilises weighted gradient orientation that is insensitive to illumination variation. Coarse and fine representation of sparse code facilitates tracking under varying scales. Subspace learning from image gradient orientation is enforced with occlusion detection during the dictionary updation stage to capture the visual characteristics of the local human appearance that supports tracking under partial occlusion with lesser tracking error. The proposed human tracking algorithm is evaluated on various datasets and shows efficient human tracking performance when compared to the other state‐of‐the‐art approaches. Furthermore, the proposed human tracking algorithm is suitable for surveillance applications.

show abstract

Compact discriminative object representation via weakly supervised learning for real‐time visual tracking

Cited by 5 publications

References 42 publications

Dictionary‐induced least squares framework for multi‐view dimensionality reduction with multi‐manifold embeddings

Dictionary‐induced least squares framework for multi‐view dimensionality reduction with multi‐manifold embeddings

On the Real Time Object Detection and Tracking

Approach to model human appearance based on sparse representation for human tracking in surveillance

Contact Info

Product

Resources

About