VeRi:大规模城市交通监控车辆重识别图像数据集

数据集 sophie ⋅ 于 3个月前 ⋅ 2079 阅读

1. VeRi dataset

为了促进车辆重识别(Re-Id)的研究,我们在现实世界城市监控场景中建立了一个名为“VeRi”的车辆Re-Id的大规模基准数据集。 VeRi的特色包括:

  • 它包含超过50,000张776辆车的图像,这些图像由20台摄像机拍摄,在24小时内覆盖1.0平方公里的面积,这使得该数据集可扩展到足以用于车辆Re-Id和其他相关研究。
  • 图像是在真实世界的无约束监视场景中捕获的,并标有不同的属性,例如: BBox,类型,颜色和品牌。因此可以学习和评估车辆Re-Id的复杂模型。
  • 每辆车在不同的视点,照明,分辨率和遮挡下由2~18台摄像机拍摄,在实际监控环境中为车辆Re-Id提供高复发率。
  • 它还标有足够的牌照和时空信息,例如板块的BBox,板条,车辆的时间戳以及相邻相机之间的距离。

filefile


2. Download

邮件发送姓名和所属机构至xinchenliu@bupt.edu.cn申请。

3. Citation

如果您使用数据集,请引用以下文章:

  • Xinchen Liu, Wu Liu, Huadong Ma, Huiyuan Fu: Large-scale vehicle re-identification in urban surveillance videos. ICME 2016: 1-6 (Best Student Paper Award, Citation=75)
  • Xinchen Liu, Wu Liu, Tao Mei, Huadong Ma: A Deep Learning-Based Approach to Progressive Vehicle Re-identification for Urban Surveillance. ECCV (2) 2016: 869-884 (Citation=56)
  • Xinchen Liu, Wu Liu, Tao Mei, Huadong Ma: PROVID: Progressive and Multimodal Vehicle Reidentification for Large-Scale Urban Surveillance. IEEE Trans. Multimedia 20(3): 645-658 (2018) (Citation=26)


4. Example codes

这里我们给出车辆搜索评估的示例代码。

在这段代码中,我们首先应该得到所有查询图像和测试图像的距离矩阵。

如在示例代码中,我们有三个距离矩阵,分别用SIFT-BOW,CN-BOW,CNN特征获得。
然后将这些矩阵与不同的权重相加以获得最终的“dist”矩阵。

然后我们从gt_image.txt和jk_image.txt中读取地面实况和垃圾图像索引(与查询图像具有相同摄像机ID的测试图像,在计算AP时不考虑它们)。

之后,对于每个查询,我们对每个测试图像的距离进行排序,并使用compute_AP函数计算Average Presicion。

最后,我们可以获得mAP,HIT @ 1,HIT @ 5和CMC曲线。

模型和距离文件可以从 BaiduPan and GoogleDrive下载.

5. State-of-the-art Results on the VeRi Dataset

Reference Year mAP Rank-1 Rank-5
[1] 2016 19.92 59.65 75.27
[2] 2016 27.77 61.44 78.78
[3] 2017 58.27 83.49 90.04
[4] 2017 57.4 86.59 92.85
[5] 2017 51.42 - -
[6] 2017 33.78 60.19 77.4
[7] 2017 60.47 85.52 95.11
[8] 2018 53.42 81.56 95.11
[9] 2018 59.47 96.24 98.97
[10] 2018 61.5 88.6 94
[11] 2018 53.53 82.9 91.6
[12] 2018 61.32 85.92 91.84
[13] 2018 53.45 83.49 92.55
[14] 2018 61.11 89.27 94.76
[15] 2018 53.35 82.06 92.31
[16] 2018 49.3 88.56 -
[17] 2018 64.78 88.62 94.52
[18] 2018 25.12 60.83 78.55
[19] 2018 60.49 77.33 88.27
[20] 2019 62.62 90.58 97.14
[21] 2019 57.44 84.39 94.05
[22] 2019 67.55 90.23 96.42
[23] 2019 61.83 88.5 94.46


Reference

[1] Liu, Xinchen, et al. "Large-scale vehicle re-identification in urban surveillance videos." ICME 2016.

[2] Liu, Xinchen, et al. "A deep learning-based approach to progressive vehicle re-identification for urban surveillance." ECCV 2016.

[3] Liu, Wu, et al. "Beyond human-level license plate super-resolution with progressive vehicle search and domain priori GAN." ACMMM 2017.

[4] Shen, Yantao, et al. "Learning deep neural networks for vehicle re-id with visual-spatio-temporal path proposals." ICCV 2017.

[5] Zhang, Yiheng, Dong Liu, and Zheng-Jun Zha. "Improving triplet-wise training of convolutional neural network for vehicle re-identification." ICME 2017.

[6] Wang, Zhongdao, et al. "Orientation invariant feature embedding and spatial temporal regularization for vehicle re-identification." ICCV 2017.

[7] Tang, Yi, et al. "Multi-modal metric learning for vehicle re-identification in traffic surveillance environment." ICIP 2017.

[8] Liu, Xinchen, et al. "PROVID: Progressive and multimodal vehicle reidentification for large-scale urban surveillance." IEEE TMM 20.3 (2018): 645-658.

[9] Bai, Yan, et al. "Group-Sensitive Triplet Embedding for Vehicle Reidentification." IEEE TMM 20.9 (2018): 2385-2399.

[10] Liu, Xiaobin, et al. "Ram: a region-aware deep model for vehicle re-identification." ICME 2018.

[11] Zhu, Jianqing, et al. "Joint feature and similarity deep learning for vehicle re-identification." IEEE Access 6 (2018): 43724-43731.

[12] Zhou, Yi, and Ling Shao. "Aware attentive multi-view inference for vehicle re-identification." CVPR 2018.

[13] Zhu, Jianqing, et al. "A shortly and densely connected convolutional neural network for vehicle re-identification." ICPR 2018.

[14] Jiang, Na, et al. "Multi-Attribute Driven Vehicle Re-Identification with Spatial-Temporal Re-Ranking." ICIP 2018.

[15] Wu, Chih-Wei, et al. "Vehicle re-identification with the space-time prior." CVPRW 2018.

[16] Kanaci, Aytac, Xiatian Zhu, and Shaogang Gong. "Vehicle Re-Identification in Context." arXiv preprint arXiv:1809.09409(2018).

[17] Wu, Fangyu, et al. "Joint Semi-supervised Learning and Re-ranking for Vehicle Re-identification." ICPR 2018.

[18] Zhou, Yi, Li Liu, and Ling Shao. "Vehicle re-identification by deep hidden multi-view inference." IEEE TIP 27.7 (2018): 3275-3287.

[19] Zhou, Yi, and Ling Shao. "Vehicle Re-Identification by Adversarial Bi-Directional LSTM Network." WACV 2018.

[20] Liu, Xinchen, et al. "PVSS: A Progressive Vehicle Search System for Video Surveillance Networks." arXiv preprint arXiv:1901.03062 (2019).

[21] Lou, Yihang, et al. "Embedding Adversarial Learning for Vehicle Re-Identification." IEEE TIP (2019).

[22] Kumar, Ratnesh, et al. "Vehicle Re-Identification: an Efficient Baseline Using Triplet Embedding." arXiv preprint arXiv:1901.01015 (2019).

[23] Zhu, Jianqing, et al. "Vehicle Re-Identification Using Quadruple Directional Deep Learning Features." IEEE TITS (2019).

微信公众号: 极市平台(ID: extrememart )
每天推送最新CV干货

本帖已被设为精华帖!
本帖由 Admin 于 2个月前 加精
回复数量: 0
暂无回复~
您需要登陆以后才能留下评论!