An improved deep learning architecture for person re-identification

By Ejaz Ahmed, Michael J. Jones, Tim K. Marks. 2015

1,262 citations recorded by the source corpus.

Subfields: Biomedical Engineering, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Biomedical Engineering

1. Yi Yang
2. Liang Zheng
3. Qi Tian
4. Shiliang Zhang
5. Zhedong Zheng
6. Shaogang Gong
7. Tao A. Xiang
8. Xiatian Zhu

In Computer Vision and Pattern Recognition

1. Yi Yang
2. Luc Van Gool
3. Xiaogang Wang
4. Dacheng Tao
5. Liang Zheng
6. Ming-Hsuan Yang
7. Qi Tian
8. Bernt Schiele

Citations in the corpus

• Set2Graph: Learning Graphs From Sets (2020, methodology): “…learning Previous models suggested for learning pairwise relations in sets were mostly of the form H 2 ( X ) = m ( φ ( x i ) , φ ( x j )) (Hsu et al., 2017; Bromley et al., 1994; Simo-Serra et al., 2015; Zagoruyko & Komodakis, 2015; Bell & Bala, 2015; Ahmed et al., 2015; Vo & Hays, 2016).”
• In Defense of the Triplet Loss for Person Re-Identification (2017, background): “However, using pretrained networks also leads to a design lock-in, and does not allow for the exploration of new deep learning advances or different architectures.”
• Multi-scale Deep Learning Architectures for Person Re-identification (2017, methodology): “(3) We improve the architecture of [1] by introducing two novel layers to implement multi-scale and saliency-based learning mechanisms.”
• Person Re-identification: Past, Present and Future (2016, methodology): “[69] improved the siamese model by computing the crossinput neighborhood difference features, which compares the features from one input image to features in neighboring”
• Learning Discriminative Features with Multiple Granularities for Person Re-Identification (2018, background): “…loss is almost the only choice of classification loss function for its strong robustness to various kinds of multi-class classification tasks, which can be used individually [1, 19, 22, 25, 36, 39, 41, 47] or combined with other losses [3, 8, 20, 43] in embedding learning procedures for Re-ID.”
• PersonNet: Person Re-identification with Deep Convolutional Neural Networks (2016, methodology): “However, little progress is witnessed in person re-id, except a few works in [1], [21], [22].”
• Deep Learning for Person Re-Identification: A Survey and Outlook (2020, background): “An improved neighborhood difference layer is proposed in [141] to capture patch feature difference and then the differences are summarized in a subsequent layer.”
• Deep Transfer Learning for Person Re-Identification (2016, methodology): “Specifically, most existing works cast the Re-ID problem as a deep metric learning problem and employ pairwise verification loss [58, 1, 43, 49, 50, 42] or triplet ranking loss [6, 29, 4], or both [52].”
• Dual Attention Matching Network for Context-Aware Feature Sequence Based Person Re-identification (2018, methodology): “For example, in [32, 3], local spatial constraints are adopted when computing spatial-patch based feature sets similarity; in [43], dense element-wise correspondences are employed when computing the distance of temporal feature sequences; in [36, 18, 1, 38], spatial correspondence structures are exp”
• Applications of Generative Adversarial Networks (GANs): An Updated Review (2019, background): “[16, 117, 118, 138] and (2) developing a distinctive signature for representing a person under different camera environment having classification typically on cross-image representation [2, 111, 124].”
• Scalable Person Re-identification on Supervised Smoothed Manifold (2017, neutral): “ImprovedDeep [1] CVPR2015 34.”
• Deep Metric Learning: A Survey (2019, background): “In [41], the authors obtained a score value for body similarity of the two images based on metric similarity.”
• Diversity Regularized Spatiotemporal Attention for Video-Based Person Re-identification (2018, methodology): “In recent years, researchers have proposed numerous deep learning based methods [1, 24, 8, 20, 44] to jointly handle both aspects.”
• I Know That Person: Generative Full Body and Face De-identification of People in Images (2017, background): “The community has been moving in leaps and bounds, making significant progress in solving very difficult problems including face recognition [25, 15], person re-identification [3, 34], image classification [11], etc.”
• Spindle Net: Person Re-identification with Human Body Region Guided Feature Decomposition and Fusion (2017, methodology): “2 (b), the 14 located body joints are assigned to three macro sets SA 1 = [1, 2, 3, 4], SA 2 = [3, 4, 5, 6, 7, 8, 9, 10], SA 3 = [9, 10, 11, 12, 13, 14] and four micro sets SB 1 = [3, 5, 6], SB 2 = [4, 7, 8], SB 3 = [9, 11, 12], SB 4 = [10, 13, 14] according to the index numbers.”
• Pedestrian Alignment Network for Large-scale Person Re-Identification (2017, methodology): “Later, [Ahmed et al., 2015] improve the performance by proposing a new part-matching layer that compares the activation of two images in neighboring pixels.”
• Improving Person Re-identification by Attribute and Identity Learning (2017, background): “Usually, the spatial constraints are integrated into the similarity learning process [1, 19, 38, 4].”
• DeepStore: In-Storage Acceleration for Intelligent Queries (2019, methodology): “The recent popular applications include person re-identification [16], style-based music retrieval [72], and question and answering systems [82].”