Pith. sign in

REVIEW 2 major objections 6 minor 45 references

Temporal Knowledge Propagation for Image-to-Video Person Re-identification

T0 review · 2 major / 6 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Propagating temporal knowledge from a video encoder into an image encoder closes much of the image-to-video person re-identification gap.

desk verdict Solid I2V Re-ID paper with large gains and strong ablations; the 'temporal knowledge' mechanism is plausible but not fully isolated from teacher capacity. read the letter →

arxiv 1908.03885 v3 pith:DSGNWDN5 submitted 2019-08-11 cs.CV

Xinqian Gu , Bingpeng Ma , Hong Chang , Shiguang Shan , Xilin Chen This is my paper · ORCID
classification cs.CV
keywords personre-identificationimage-to-videoretrievaltemporalknowledgepropagationdistillationnon-localnetworksvideosurveillancefeatureembedding
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper claims that the performance gap in image-to-video person re-identification stems from the query image lacking the temporal information that gallery videos carry, and that this gap can be largely closed at training time. Its Temporal Knowledge Propagation (TKP) method makes a plain image encoder mimic the per-frame features of a video encoder that contains non-local temporal blocks, using two complementary losses in a shared feature space. On the MARS dataset, TKP raises top-1 accuracy from 67.1 percent to 75.6 percent and mAP from 55.5 to 65.1 percent, and it also improves image-to-image retrieval, showing the image features themselves become more robust. A sympathetic reader would care because the method offers a general way to transfer a temporal capability from one network to another without changing the test-time architecture.

What carries the argument

The load-bearing object is the TKP loss $L_{TKP}=L^F_{TKP}+L^D_{TKP}$. Here $L^F_{TKP}$ is the mean squared error between image features $i_{nt}$ and video frame features $f_{nt}$ for every frame $t$ of every clip $n$, and $L^D_{TKP}$ is the squared Frobenius norm of the difference between the image distance matrix $D^{img}\in\mathbb{R}^{NT\times NT}$ and the video frame distance matrix $D^{vid}\in\mathbb{R}^{NT\times NT}$. These losses are added to classification and integrated triplet losses, and they carry the temporal knowledge from the video representation network (ResNet-50 with non-local blocks) to the image representation network (plain ResNet-50), while the TKP gradient is blocked from updating the video network.

What would settle it

Train TKP on MARS, then test with query images taken from a different camera or from later video frames rather than the first frame; if top-1 accuracy drops back to the no-transfer baseline (or below the 8.5-point improvement), the transferred temporal knowledge does not generalize beyond the training video distribution.

Watch

Extended reading notes

Core claim

The central discovery is that temporal knowledge—resilience to occlusion, blur, and pose variation that a video model acquires by seeing frames in relation to one another—can be transferred into a single-image feature extractor. The transfer is achieved by the TKP loss, which combines a feature-level mean squared error between image features and corresponding video frame features with a distance-level Frobenius norm between the cross-sample Euclidean distance matrices of the two modalities. Both image and video networks are ResNet-50 based; the video network adds non-local blocks, and during training the TKP loss is not back-propagated through the video network, preserving its temporal modeling ability. The result is that image features pay attention to more of the person foreground and the image-video feature distributions align, which directly improves image-to-video retrieval.

Load-bearing premise

The method assumes that the temporal knowledge a video network learns from training frames can be transferred to a still-image encoder by fitting its outputs, and that this transfer holds for query images that may come from different cameras, poses, or image quality than the training frames.

Editorial extensions

If this is right

  • I2V Re-ID systems can use a single still image as query while the gallery remains video, with accuracy approaching video-to-video levels.
  • The transfer improves I2I Re-ID as well (MARS mAP from 49.2 to 55.0), so the robustness gain lives in the image features, not just the matching protocol.
  • Any pair of teacher/student architectures that differ in temporal modeling can adopt the same two-term transfer loss.
  • The method achieves state-of-the-art I2V results on MARS (top-1 75.6, mAP 65.1) and Duke (top-1 77.9, mAP 75.9) without changing test-time computation.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The success of combining feature and distance distillation suggests a general recipe for cross-modal capability transfer: match both the raw outputs and the relational structure of the embedding space; the same recipe could apply to transferring temporal knowledge to single-image action recognition or person search from a single frame.
  • Because adding TKP to the baseline also lifts I2I performance, the method indirectly indicates that a portion of the I2V gap is feature-asymmetry rather than missing information; an adversarial or normalization-based alignment might recover part of the same gain.
  • A natural extension is to train the image network on frames from multiple camera domains while keeping the video network fixed, then test whether the transferred robustness survives a domain shift in the query images.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

2 major / 6 minor

Summary. The paper proposes Temporal Knowledge Propagation (TKP) for image-to-video person re-identification (I2V Re-ID). The method trains an image representation network (ResNet-50) and a video representation network (ResNet-50 with non-local blocks) jointly. The TKP loss has two terms: a feature-matching term (Eq. 3, MSE between image frame features and video frame features) and a distance-matching term (Eq. 4, matching cross-sample distance matrices), with gradients of both terms blocked from the video network. These are combined with classification and integrated triplet losses. At test time, query image features come from the image network and gallery video features from the video network. Experiments on MARS, Duke, and iLIDS-VID report large gains over a baseline (MARS: 67.1% to 75.6% top-1, 55.5% to 65.1% mAP) and state-of-the-art comparisons. Extensive ablations cover the non-local blocks, gradient blocking, a pre-trained teacher, triplet-loss variants, and clip length T.

Significance. If the mechanism claim is accepted, the paper is a solid empirical contribution: it reports consistent, substantial improvements over a strong baseline on two multi-camera datasets, includes a well-designed gradient-blocking analysis, and releases code. The ablation suite is unusually thorough for this area, and the I2I/V2V comparisons show that the image features improve without degrading the video features. The main weakness is that the central conceptual attribution of the gains to 'temporal knowledge' is not fully isolated from capacity effects, because the non-local teacher differs from the no-non-local teacher in both temporal modeling and parameter count. The benchmark numbers themselves are credible and likely reproducible, but the distinct contribution of temporal modeling needs one or two additional control experiments before the paper's core claim can be considered established.

major comments (2)
  1. [Section 4.8, Table 6] The central claim that the additional gain enabled by non-local blocks is due to temporal knowledge is confounded by capacity and by a change in the distillation target. In the no-non-local rows, the teacher target is the temporally averaged video feature v_n, not the per-frame feature f_nt used in Eq. (3), and the teacher also has fewer parameters. The 6.1-point top-1 gap between TKP-F without non-local blocks (68.9) and TKP-F with non-local blocks (75.0) could therefore come from the higher-capacity teacher, from the different target features, or from genuine temporal relations. A control using a capacity-matched teacher that is temporally blind (e.g., replacing non-local blocks with equal-capacity residual blocks that process each frame independently, or training the video network on temporally shuffled frames) is needed to attribute the gain to temporal modeling.
  2. [Section 4.3, Table 2] All MARS and Duke results are reported as single runs. The differences between TKP-F, TKP-D, and TKP are moderate (e.g., 75.0 vs. 75.6 top-1 on MARS; 63.1 vs. 64.2 vs. 65.1 mAP), and without run-to-run variance it is not possible to judge whether the complementarity of the two TKP losses is a genuine effect or stochastic noise. The authors should report the mean and standard deviation over at least three training runs, or provide a significance test, especially because the claim that the two transfer ways are 'complementary' in Section 4.3 rests on these small margins.
minor comments (6)
  1. [Section 3.3] There is a typo in the sentence 'in our emperiments, LF_TKP and LF_TKP are not back-propagated': 'emperiments' should be 'experiments', and the second loss symbol should be LD_TKP rather than LF_TKP.
  2. [Section 4.1 and 4.2] The iLIDS-VID protocol description is ambiguous: the text says only the first frames of videos from the first camera are used for training and testing, but TKP training requires video clips of T frames. Please clarify how the video clips are formed for the iLIDS-VID fine-tuning stage and how the gallery videos are used at test time.
  3. [Section 4.2] The statement about sampling '4 frames with a stride of 8 frames' and duplicating videos shorter than 32 frames is clear, but the choice of stride 8 and the duplication rule deserve a sentence of justification because they affect the temporal span of the clips and therefore the temporal knowledge available to the teacher.
  4. [Section 5] In the conclusion, 'start-of-the-art' should be 'state-of-the-art'.
  5. [Figure 2] The figure caption and legend refer to 'the process of TKP via features' and 'TKP via cross sample distances', which is helpful, but the red arrow for back-propagation should be labeled to indicate that it stops at the image network and does not enter the video network, to match the description in Section 3.3.
  6. [Section 4.5, Figure 4] In the discussion of the Duke gap, the text attributes the larger I2V-to-V2V gap to longer average video length. This is plausible, but the explanation would be strengthened by reporting the actual loss of information, e.g., the fraction of gallery videos where the first frame is occluded or blurred.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the TKP losses are training objectives and the reported Re-ID gains are independently measured empirical outcomes.

full rationale

The paper's derivation chain is a standard supervised training setup: define image features (Eq. 1) and video frame features (Eq. 2), define temporal knowledge propagation losses in terms of feature matching (Eq. 3) and cross-sample distance matching (Eq. 4), combine them with classification and triplet losses (Eq. 10), train on video clips, and then evaluate on held-out image-to-video retrieval. The reported accuracies in Table 2 are measured after training; they are not fed back into any equation as inputs. The TKP loss is an objective function, not a fitted parameter that is later renamed as a prediction. The claim that temporal knowledge is transferred is an interpretation of the distillation-style loss, and the ablations in Tables 6 and 7 provide empirical comparisons rather than circular justifications. There is no self-citation chain that supplies a load-bearing premise, no imported uniqueness theorem, and no ansatz smuggled in via citation to prior work. Even if the attribution of the improvement specifically to temporal modeling is underdetermined by the ablations, that is a question of experimental validity or confound control, not circularity. The derivation does not assume the results it claims to produce.

Assumptions & free parameters 3 free parameters · 3 assumptions · 0 invented entities

The central claim rests on standard deep learning components and one task-specific assumption about distribution transfer. The free parameters are hyperparameters selected by hand or small ablations, not quantities fitted to make the main results. No new entities are postulated.

free parameters (3)
  • Video clip length T = 4
    Selected via ablation on MARS (Figure 7); top-1 and mAP are highest at T=4, used for all datasets.
  • Batch sampling parameters P and K = P=4, K=4
    Set by hand in Section 4.2; no ablation or sensitivity analysis is reported.
  • Triplet loss margin m = not reported
    Pre-defined margin in Eqs. (6)-(9); the paper does not state its value or how it was chosen.
assumptions (3)
  • domain assumption Query images are drawn from the same distribution as training video frames (first-frame query protocol).
    Section 4.1 defines the I2V protocol using the first frame of each video as query; TKP transfers knowledge from training video frames, so this distribution match is load-bearing.
  • domain assumption The video network's non-local blocks learn temporal knowledge that improves feature robustness, and this knowledge is captured in per-frame features f_nt.
    The TKP losses in Eqs. (3)-(4) use f_nt as targets; Table 6 shows non-local blocks help, but the causal role of temporal knowledge is inferred, not directly measured.
  • ad hoc to paper Gradients of the TKP loss must not reach the video network, or the non-local weights collapse to zero and temporal modeling is lost.
    Section 3.3 explains the trivial solution and the authors block gradients; Table 7 empirically supports this, but it is a design choice specific to this method.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Temporal Knowledge Propagation for Image-to-Video Person Re-identification." pith.science (2026). https://pith.science/paper/DSGNWDN5

@misc{pith2026190803885,
  author       = {Pith},
  title        = {Pith review of: Temporal Knowledge Propagation for Image-to-Video Person Re-identification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DSGNWDN5}},
  note         = {Machine review of arXiv:1908.03885}
}
read the original abstract

In many scenarios of Person Re-identification (Re-ID), the gallery set consists of lots of surveillance videos and the query is just an image, thus Re-ID has to be conducted between image and videos. Compared with videos, still person images lack temporal information. Besides, the information asymmetry between image and video features increases the difficulty in matching images and videos. To solve this problem, we propose a novel Temporal Knowledge Propagation (TKP) method which propagates the temporal knowledge learned by the video representation network to the image representation network. Specifically, given the input videos, we enforce the image representation network to fit the outputs of video representation network in a shared feature space. With back propagation, temporal knowledge can be transferred to enhance the image features and the information asymmetry problem can be alleviated. With additional classification and integrated triplet losses, our model can learn expressive and discriminative image and video features for image-to-video re-identification. Extensive experiments demonstrate the effectiveness of our method and the overall results on two widely used datasets surpass the state-of-the-art methods by a large margin. Code is available at: https://github.com/guxinqian/TKP

Figures

Figures reproduced from arXiv: 1908.03885 by the authors.

Figure 1
Figure 1. Visualisation of the feature maps. Due to lack of tem [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The framework of TKP method in I2V Re-ID training. SAP and TAP represent spatial average pooling and temporal average [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. The pipeline of I2V Re-ID testing. 4. Experiments 4.1. Datasets and Evaluation Protocol Datasets. We evaluate our method on MARS [41], DukeMTMC-VideoReID (Duke) [34] and iLIDS-VID [32] datasets. Among them, MARS and Duke are multi-camera datasets, while iLIDS-VID is captured by only two cam￾eras. The amounts of person videos on MARS, Duke and iLIDS-VID are 20478, 5534 and 600 respectively, and the average lengths of… view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Comparison among I2I, I2V and V2V Re-ID on the [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: The visualisation of feature maps before/after TKP on [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: The visualisation of feature distribution before/after [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: The results with different T on the MARS dataset. representation network? As discussed in Section 3.3, en￾forcing TKP loss to propagate gradient to video represen￾taiton network will degenerate video representations w.r.t. temporal knowledge. To verify this, we add two…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

45 extracted references · 32 canonical work pages

  1. [1]

    Scalable person re- identification on supervised smoothed manifold

    Song Bai, Xiang Bai, and Qi Tian. Scalable person re- identification on supervised smoothed manifold. In CVPR, 2017

    work page 2017

  2. [2]

    Model compression

    Cristian Bucilu, Rich Caruana, and Alexandru Niculescu- Mizil. Model compression. In KDD, 2006

    work page 2006

  3. [3]

    Video person re-identification with compet- itive snippet-similarity aggregation and co-attentive snippet embedding

    Dapeng Chen, Hongsheng Li, Tong Xiao, Shuai Yi, and Xi- aogang Wang. Video person re-identification with compet- itive snippet-similarity aggregation and co-attentive snippet embedding. In CVPR, 2018

    2018

  4. [4]

    Dark- rank: Accelerating deep metric learning via cross sample similarities transfer

    Yuntao Chen, Naiyan Wang, and Zhaoxiang Zhang. Dark- rank: Accelerating deep metric learning via cross sample similarities transfer. In AAAI, 2018

    work page 2018

  5. [5]

    Dahjung Chung, Khalid Tahboub, and Edward J. Delp. A two stream siamese convolutional neural network for person re-identification. In ICCV, 2017

    work page 2017

  6. [6]

    Learning spatiotemporal features with 3d convolutional networks

    Tran Du, Lubomir Bourdev, Rob Fergus, Lorenzo Torresani, and Manohar Paluri. Learning spatiotemporal features with 3d convolutional networks. In ICCV, 2015

    work page 2015

  7. [7]

    Christoph Feichtenhofer, Axel Pinz, and Richard P. Wildes. Spatiotemporal multiplier networks for video action recog- nition. In CVPR, 2017

    work page 2017

  8. [8]

    Deep multi-metric learning for person re- identification

    Yongxin Ge, Xinqian Gu, Min Chen, Hongxing Wang, and Dan Yang. Deep multi-metric learning for person re- identification. In ICME, 2018

    work page 2018

Show all 45 references
  1. [9]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In CVPR, 2016

    2016

  2. [10]

    In defense of the triplet loss for person re-identification

    Alexander Hermans, Lucas Beyer, and Bastian Leibe. In defense of the triplet loss for person re-identification. ArXiv:1703.07737, 2017

    2017 arXiv

  3. [11]

    Distilling the knowledge in a neural network

    Geoffrey Hinton, Oriol Vinyals, and Jeff Dean. Distilling the knowledge in a neural network. In NIPS workshop, 2014

    2014

  4. [12]

    Interaction-and-aggregation network for person re-identification

    Ruibing Hou, Bingpeng Ma, Hong Chang, Xinqian Gu, Shiguang Shan, and Xilin Chen. Interaction-and-aggregation network for person re-identification. In CVPR, 2019

    2019

  5. [13]

    Vrstc: Occlusion-free video person re-identification

    Ruibing Hou, Bingpeng Ma, Hong Chang, Xinqian Gu, Shiguang Shan, and Xilin Chen. Vrstc: Occlusion-free video person re-identification. In CVPR, 2019

    2019

  6. [14]

    Learning euclidean-to-riemannian metric for point- to-set classification

    Zhiwu Huang, Ruiping Wang, Shiguang Shan, and Xilin Chen. Learning euclidean-to-riemannian metric for point- to-set classification. In CVPR, 2014

    2014

  7. [15]

    3d convo- lutional neural networks for human action recognition

    Shuiwang Ji, Wei Xu, Ming Yang, and Kai Yu. 3d convo- lutional neural networks for human action recognition. In ICML, 2010

    2010

  8. [16]

    Adam: A method for stochastic optimization

    Diederik P Kingma and Jimmy Ba. Adam: A method for stochastic optimization. In ICLR, 2015

    2015

  9. [17]

    Joseph B. Kruskal. Multidimensional scaling by optimizing goodness of fit to a nonmetric hypothesis. Psychometrika, 1964

    1964

  10. [18]

    Large scale metric learning from equivalence constraints

    Martin Kstinger, Martin Hirzer, Paul Wohlhart, Peter M Roth, and Horst Bischof. Large scale metric learning from equivalence constraints. In CVPR, 2012

    2012

  11. [19]

    The approximation power of moving least- squares

    David Levin. The approximation power of moving least- squares. Mathematics of Computation, 1998

    1998

  12. [20]

    Diversity regularized spatiotemporal attention for video- based person re-identification

    Shuang Li, Slawomir Bak, Peter Carr, and Xiaogang Wang. Diversity regularized spatiotemporal attention for video- based person re-identification. In CVPR, 2018

    2018

  13. [21]

    Shengcai Liao, Yang Hu, Xiangyu Zhu, and Stan Z. Li. Per- son re-identification by local maximal occurrence represen- tation and metric learning. In CVPR, 2015

    2015

  14. [22]

    Visualizing data using t-sne

    Laurens van der Maaten and Geoffrey Hinton. Visualizing data using t-sne. JMLR, 2008

    2008

  15. [23]

    Recurrent convolutional network for video-based per- son re-identification

    Niall Mclaughlin, Jesus Martinez Del Rincon, and Paul Miller. Recurrent convolutional network for video-based per- son re-identification. In CVPR, 2016

    2016

  16. [24]

    Beyond short snippets: Deep networks for video classification

    Yue Hei Ng, Matthew Hausknecht, Sudheendra Vijaya- narasimhan, Oriol Vinyals, Rajat Monga, and George Toderici. Beyond short snippets: Deep networks for video classification. In CVPR, 2015

    2015

  17. [25]

    Fitnets: Hints for thin deep nets

    Adriana Romero, Nicolas Ballas, Samira Ebrahimi Kahou, Antoine Chassang, Carlo Gatta, and Yoshua Bengio. Fitnets: Hints for thin deep nets. In ICLR, 2015

    2015

  18. [26]

    Imagenet large scale visual recognition challenge

    Olga Russakovsky, Jia Deng, Hao Su, Jonathan Krause, San- jeev Satheesh, Sean Ma, Zhiheng Huang, Andrej Karpathy, Aditya Khosla, and Michael Bernstein. Imagenet large scale visual recognition challenge. IJCV, 2015

    2015

  19. [27]

    Image de- formation using moving least squares

    Scott Schaefer, Travis McPhail, and Joe Warren. Image de- formation using moving least squares. In SIGGRAPH, 2006

    2006

  20. [28]

    Kot, and Gang Wang

    Jianlou Si, Honggang Zhang, Chun-Guang Li, Jason Kuen, Xiangfei Kong, Alex C. Kot, and Gang Wang. Dual attention matching network for context-aware feature sequence based person re-identification. In CVPR, 2018

    2018

  21. [29]

    Mask-guided contrastive attention model for person re-identification

    Chunfeng Song, Yan Huang, Wanli Ouyang, and Liang Wang. Mask-guided contrastive attention model for person re-identification. In CVPR, 2018

    2018

  22. [30]

    Beyond part models: Person retrieval with refined part pooling (and a strong convolutional baseline)

    Yifan Sun, Liang Zheng, Yi Yang, Qi Tian, and Shengjin Wang. Beyond part models: Person retrieval with refined part pooling (and a strong convolutional baseline). InECCV, 2018

    2018

  23. [31]

    P2snet : Can an image match a video for person re-identification in an end-to-end way? TCSVT, 2017

    Guangcong Wang, Jianhuang Lai, and Xiaohua Xie. P2snet : Can an image match a video for person re-identification in an end-to-end way? TCSVT, 2017

    2017

  24. [32]

    Person re-identification by video ranking

    Taiqing Wang, Shaogang Gong, Xiatian Zhu, and Shengjin Wang. Person re-identification by video ranking. In ECCV, 2014

    2014

  25. [33]

    Non-local neural networks

    Xiaolong Wang, Ross Girshick, Abhinav Gupta, and Kaim- ing He. Non-local neural networks. In CVPR, 2018

    2018

  26. [34]

    Exploit the unknown gradually: One-shot video-based person re-identification by stepwise learning

    Yu Wu, Yutian Lin, Xuanyi Dong, Yan Yan, Wanli Ouyang, and Yi Yang. Exploit the unknown gradually: One-shot video-based person re-identification by stepwise learning. In CVPR, 2018

    2018

  27. [35]

    Top- push video-based person re-identification

    Jinjie You, Ancong Wu, Xiang Li, and Wei-Shi Zheng. Top- push video-based person re-identification. In CVPR, 2016

    2016

  28. [36]

    Hard-aware point-to-set deep metric for person re-identification

    Rui Yu, Zhiyong Dou, Song Bai, Zhaoxiang Zhang, Yongchao Xu, and Xiang Bai. Hard-aware point-to-set deep metric for person re-identification. In ECCV, 2018

    2018

  29. [37]

    Image-to-video per- son re-identification with temporally memorized similarity learning

    Dongyu Zhang, Wenxi Wu, Hui Cheng, Ruimao Zhang, Zhenjiang Dong, and Zhaoquan Cai. Image-to-video per- son re-identification with temporally memorized similarity learning. TCSVT, 2017

    2017

  30. [38]

    Multi-shot pedestrian re-identification via sequential decision making

    Jianfu Zhang, Naiyan Wang, and Liqing Zhang. Multi-shot pedestrian re-identification via sequential decision making. In CVPR, 2018

    2018

  31. [39]

    Aligne- dreid: Surpassing human-level performance in person re- identification

    Xuan Zhang, Hao Luo, Xing Fan, Weilai Xiang, Yixiao Sun, Qiqi Xiao, Wei Jiang, Chi Zhang, and Jian Sun. Aligne- dreid: Surpassing human-level performance in person re- identification. ArXiv:1711.08184, 2017

    2017 arXiv

  32. [40]

    Hospedales, and Huchuan Lu

    Ying Zhang, Tao Xiang, Timothy M. Hospedales, and Huchuan Lu. Deep mutual learning. In CVPR, 2018

    2018

  33. [41]

    Mars: A video benchmark for large-scale person re-identification

    Liang Zheng, Zhi Bie, Yifan Sun, Jingdong Wang, Chi Su, Shengjin Wang, and Qi Tian. Mars: A video benchmark for large-scale person re-identification. In ECCV, 2016

    2016

  34. [42]

    Scalable person re-identification: A benchmark

    Liang Zheng, Liyue Shen, Lu Tian, Shengjin Wang, Jing- dong Wang, and Qi Tian. Scalable person re-identification: A benchmark. In ICCV, 2015

    2015

  35. [43]

    From point to set: Extend the learning of distance metrics

    Pengfei Zhu, Lei Zhang, Wangmeng Zuo, and David Zhang. From point to set: Extend the learning of distance metrics. In ICCV, 2013

    2013

  36. [44]

    Learning hetero- geneous dictionary pair with feature projection matrix for pedestrian video retrieval via single query image

    Xiaoke Zhu, Xiao-Yuan Jing, Fei Wu, Yunhong Wang, Wangmeng Zuo, and Wei-Shi Zheng. Learning hetero- geneous dictionary pair with feature projection matrix for pedestrian video retrieval via single query image. In AAAI, 2017

    2017

  37. [45]

    Image to video person re-identification by learning heterogeneous dictionary pair with feature projection matrix

    Xiaoke Zhu, Xiao-Yuan Jing, Xinge You, Wangmeng Zuo, Shiguang Shan, and Wei-Shi Zheng. Image to video person re-identification by learning heterogeneous dictionary pair with feature projection matrix. TIFS, 2018

    2018

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.