Learning Modal-Invariant Angular Metric by Cyclic Projection Network for VIS-NIR Person Re-Identification

Zhang, Quan; Lai, Jianhuang; Xie, Xiaohua

doi:10.1109/tip.2021.3112035

Cited by 34 publications

(7 citation statements)

References 54 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Non-Generative Cross-Modal Re-ID: Non-generative cross-modal Re-ID approaches aim to find discriminative feature embedding by either deploying feature learning approaches [ 67 , 68 , 69 , 70 , 71 , 72 , 73 , 74 , 75 , 76 , 77 , 78 , 79 , 80 , 81 , 82 , 83 , 84 , 85 , 86 , 87 ] or utilizing metric learning approaches. Most of the existing feature learning approaches focus on extracting global features by employing: one individual branch each for visible and infrared images [ 67 ]; a cross-modality shared-specific feature transfer algorithm [ 68 ]; consistency at the feature and classifier levels [ 69 ]; or attention mechanisms [ 70 , 71 ].…”

Section: Cross-modal Person Re-identificationmentioning

confidence: 99%

“…The aim of metric learning-based Re-ID approaches [ 82 , 83 , 84 , 85 , 86 ] is to direct feature representations to fulfill certain objective functions for better recognition. Owing to the metric learning approach, a two-stream network architecture was proposed in [ 82 ], in which contrastive loss is utilized to bridge the gap between two modalities and enhance the modality invariance of learned representations.…”

Section: Cross-modal Person Re-identificationmentioning

confidence: 99%

“…Ye et al [ 84 ] proposed a triplet loss that uses the angles between instances as a distance measure for optimizing the network. Similarly, in [ 86 ], a feature space is mapped onto an angular space and utilized for triplet loss and center loss for network optimization. In [ 87 ], a modified version called quadruplet loss is proposed, which combines triple loss with classification loss to optimize the proposed hyperspherical manifold-embedded network.…”

Section: Cross-modal Person Re-identificationmentioning

confidence: 99%

“…In recent years, Re-ID researchers have focused on cross-modal matching and have proposed several methods. Almost all methods [ 67 , 68 , 69 , 70 , 71 , 72 , 73 , 74 , 75 , 76 , 77 , 78 , 79 , 80 , 81 , 82 , 83 , 84 , 85 , 86 , 87 , 88 , 89 , 90 , 91 , 92 , 93 , 94 , 95 , 96 , 97 , 98 ] of RGB–IR cross-modal Re-IDs used two benchmark datasets: SYSU-MM01 [ 45 ] and RegDB [ 46 ]. To perform experiments in their proposed methods, they used two RGB–IR benchmark datasets: SYSU-MM01 [ 45 ] and RegDB [ 46 ].…”

Section: Datasetsmentioning

confidence: 99%

See 3 more Smart Citations

Person Re-Identification with RGB–D and RGB–IR Sensors: A Comprehensive Survey

Uddin

Bhuiyan

Bappee

et al. 2023

Sensors

View full text Add to dashboard Cite

Learning about appearance embedding is of great importance for a variety of different computer-vision applications, which has prompted a surge in person re-identification (Re-ID) papers. The aim of these papers has been to identify an individual over a set of non-overlapping cameras. Despite recent advances in RGB–RGB Re-ID approaches with deep-learning architectures, the approach fails to consistently work well when there are low resolutions in dark conditions. The introduction of different sensors (i.e., RGB–D and infrared (IR)) enables the capture of appearances even in dark conditions. Recently, a lot of research has been dedicated to addressing the issue of finding appearance embedding in dark conditions using different advanced camera sensors. In this paper, we give a comprehensive overview of existing Re-ID approaches that utilize the additional information from different sensor-based methods to address the constraints faced by RGB camera-based person Re-ID systems. Although there are a number of survey papers that consider either the RGB–RGB or Visible-IR scenarios, there are none that consider both RGB–D and RGB–IR. In this paper, we present a detailed taxonomy of the existing approaches along with the existing RGB–D and RGB–IR person Re-ID datasets. Then, we summarize the performance of state-of-the-art methods on several representative RGB–D and RGB–IR datasets. Finally, future directions and current issues are considered for improving the different sensor-based person Re-ID systems.

show abstract

Section: Cross-modal Person Re-identificationmentioning

confidence: 99%

Section: Cross-modal Person Re-identificationmentioning

confidence: 99%

Section: Cross-modal Person Re-identificationmentioning

confidence: 99%

Section: Datasetsmentioning

confidence: 99%

See 2 more Smart Citations

Person Re-Identification with RGB–D and RGB–IR Sensors: A Comprehensive Survey

Uddin

Bhuiyan

Bappee

et al. 2023

Sensors

View full text Add to dashboard Cite

show abstract

“…Cai et al [43] proposed a dual-modality hard mining triplet-center loss (DTCL) which can reduce computational cost and mine hard triplet samples. In order to eliminate the effect of inconsistent feature distribution in different modalities, Zhang et al [44] mapped the feature space to angular space and proposed several loss functions to conduct specific angular metric learning.…”

Section: Metric Learningmentioning

confidence: 99%

Visible-Infrared Person Re-Identification: A Comprehensive Survey and a New Setting

et al. 2022

View full text Add to dashboard Cite

Person re-identification (ReID) plays a crucial role in video surveillance with the aim to search a specific person across disjoint cameras, and it has progressed notably in recent years. However, visible cameras may not be able to record enough information about the pedestrian’s appearance under the condition of low illumination. On the contrary, thermal infrared images can significantly mitigate this issue. To this end, combining visible images with infrared images is a natural trend, and are considerably heterogeneous modalities. Some attempts have recently been contributed to visible-infrared person re-identification (VI-ReID). This paper provides a complete overview of current VI-ReID approaches that employ deep learning algorithms. To align with the practical application scenarios, we first propose a new testing setting and systematically evaluate state-of-the-art methods based on our new setting. Then, we compare ReID with VI-ReID in three aspects, including data composition, challenges, and performance. According to the summary of previous work, we classify the existing methods into two categories. Additionally, we elaborate on frequently used datasets and metrics for performance evaluation. We give insights on the historical development and conclude the limitations of off-the-shelf methods. We finally discuss the future directions of VI-ReID that the community should further address.

show abstract