Joint identification–verification for person re-identification: A four stream deep learning approach with improved quartet loss function

Khatun, Amena; Denman, Simon; Sridharan, Sridha; Fookes, Clinton

doi:10.1016/j.cviu.2020.102989

Cited by 22 publications

(16 citation statements)

References 1 publication

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The pose attention-guided appearance network and the pose attention-guided generation network progressively model the appearance and shape of a person to synthesise a person image with the target pose, while keeping the appearance and identity constant. The discriminative re-ID module is trained with the quartet loss function [18] to boost re-ID performance. As shown in Figure 2, the condition image, I c , and the condition pose, P c , are first fed into the appearance encoder, E A , and pose encoder, E p , to generate the appearance map, f I o , and the pose map, f P o .…”

Section: Proposed Methodsmentioning

confidence: 99%

“…However, images from the same identity may lie close to each other in feature space, but with a large intra-class distance as [35] does not specify the distance between the positive pair. Thus, [18] proposed a new loss function that uses a quartet of images and minimises the distance between the positive pair more than the distance between the negative pair regardless of whether the probe image comes from the same identity or not, and simultaneously ensures that the intra-class features will be close to each other.…”

Section: Deep Learning For Re-idmentioning

confidence: 99%

“…We also ensure to preserve the identity of a person during pose translation using the semantic-consistency loss. Furthermore, for re-ID, we adopted the quartet loss [18] to minimise the intra-class distance more than the inter-class distance in feature space.…”

Section: Pose Variation and Re-idmentioning

confidence: 99%

“…For re-identification, we adopt the improved quartet loss function for verification [18] that takes four images as input. For the improved quartet loss, the input images are represented as,…”

Section: Discriminative Re-id Networkmentioning

confidence: 99%

“…The newly generated pose-transferred images help to enhance the discriminative ability of the re-identification model, and thus increase re-ID performance. For re-ID, we adopted our earlier proposed improved quartet loss for verification [18] and softmax identification loss. The major contributions of this paper can be summarised as follows:…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Pose-driven Attention-guided Image Generation for Person Re-Identification

Khatun¹,

Denman²,

Sridharan³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Person re-identification (re-ID) concerns the matching of subject images across different camera views in a multi camera surveillance system. One of the major challenges in person re-ID is pose variations across the camera network, which significantly affects the appearance of a person. Existing development data lack adequate pose variations to carry out effective training of person re-ID systems. To solve this issue, in this paper we propose an end-to-end pose-driven attention-guided generative adversarial network, to generate multiple poses of a person. We propose to attentively learn and transfer the subject pose through an attention mechanism. A semantic-consistency loss is proposed to preserve the semantic information of the person during pose transfer. To ensure fine image details are realistic after pose translation, an appearance discriminator is used while a pose discriminator is used to ensure the pose of the transferred images will exactly be the same as the target pose. We show that by incorporating the proposed approach in a person re-identification framework, realistic pose transferred images and state-of-the-art re-identification results can be achieved.

show abstract

Section: Proposed Methodsmentioning

confidence: 99%

Section: Deep Learning For Re-idmentioning

confidence: 99%

Section: Pose Variation and Re-idmentioning

confidence: 99%

Section: Discriminative Re-id Networkmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Pose-driven Attention-guided Image Generation for Person Re-Identification

Khatun¹,

Denman²,

Sridharan³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

Quantum Transfer Learning with Adversarial Robustness for Classification of High‐Resolution Image Datasets

Khatun,

Usman

2024

Adv Quantum Tech

View full text Add to dashboard Cite

The application of quantum machine learning to large‐scale high‐resolution image datasets is not yet possible due to the limited number of qubits and relatively high level of noise in the current generation of quantum devices. In this work, this challenge is addressed by proposing a quantum transfer learning (QTL) architecture that integrates quantum variational circuits with a classical machine learning network pre‐trained on ImageNet dataset. Through a systematic set of simulations over a variety of image datasets such as Ants & Bees, CIFAR‐10, and Road Sign Detection, the superior performance of the QTL approach over classical and quantum machine learning without involving transfer learning is demonstrated. Furthermore, the adversarial robustness of QTL architecture with and without adversarial training is evaluated, confirming that our QTL method is adversarially robust against data manipulation attacks and outperforms classical methods.

show abstract

Collaborative learning mutual network for domain adaptation in person re-identification

Tay

Yap

2022

Neural Comput & Applic

View full text Add to dashboard Cite

Joint identification–verification for person re-identification: A four stream deep learning approach with improved quartet loss function

Cited by 22 publications

References 1 publication

Pose-driven Attention-guided Image Generation for Person Re-Identification

Pose-driven Attention-guided Image Generation for Person Re-Identification

Quantum Transfer Learning with Adversarial Robustness for Classification of High‐Resolution Image Datasets

Collaborative learning mutual network for domain adaptation in person re-identification

Contact Info

Product

Resources

About