Performance Measures and a Data Set for Multi-target, Multi-camera Tracking

By Ergys Ristani, Francesco Solera, Roger S. Zou, Rita Cucchiara, Carlo Tomasi. 2016

3,097 citations recorded by the source corpus.

To help accelerate progress in multi-target, multi-camera tracking systems, we present (i) a new pair of precision-recall measures of performance that treats errors of all types uniformly and emphasizes correct identification over sources of error; (ii) the largest fully-annotated and calibrated data set to date with more than 2 million frames of 1080 p, 60 fps video taken by 8 cameras observing more than 2,700 identities over 85 min; and (iii) a reference software system as a comparison baseline. We show that (i) our measures properly account for bottom-line identity match performance in the multi-camera setting; (ii) our data set poses realistic challenges to current trackers; and (iii) the performance of our system is comparable to the state of the art.

Subfields: Aerospace 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 Aerospace Engineering

In Computer Vision and Pattern Recognition

Citations in the corpus

Sequential Attend, Infer, Repeat: Generative Modelling of Moving Objects (2018, methodology): “To evaluate the model in a more challenging, real-world setting, we turn to data from static CCTV cameras of the DukeMTMC dataset (Ristani et al., 2016).”
TrackFormer: Multi-Object Tracking with Transformers (2021, background): “While our gain over CenterTrack seems similar across the two filtering techniques for MOTA (see Table A.1), the gain in IDF1 is significantly larger under the more challenging IoU-based protocol, which suggests that CenterTrack benefits from the less challenging CD-based filtering protocol, while Tr”
Mutual Mean-Teaching: Pseudo Label Refinery for Unsupervised Domain Adaptation on Person Re-identification (2020, methodology): “We utilize k-means clustering algorithm and the number M t of pseudo classes is set as 500, 700, 900 for Market-1501 and DukeMTMC-reID, and 500, 1000, 1500, 2000 (Zheng et al., 2015), DukeMTMC-reID (Ristani et al., 2016), and MSMT17 (Wei et al., 2018) datasets, where MMT-M t represents the result wi”
Person Transfer GAN to Bridge Domain Gap for Person Re-identification (2017, neutral): “Dataset MSMT17 Duke [41, 27] Market [39] CUHK03 [20] CUHK01 [19] VIPeR [8] PRID [10] CAVIAR [3]”
A Pose-Sensitive Embedding for Person Re-identification with Expanded Cross Neighborhood Re-ranking (2017, methodology): “The Duke-MTMC-reID (Duke) dataset is created from data of eight cameras.”
Hierarchical Clustering With Hard-Batch Triplet Loss for Person Re-Identification (2019, background): “DukeMTMC-reID DukeMTMC-reID [34] is a sub-set of pedestrian re-identiﬁcation dataset DukeMTMC [17].”
Vision Meets Drones: A Challenge (2018, background): “Ristani et al. [27] organize a challenge and present a large-scale fully-annotated and calibrated dataset, including more than 2 million 1080p video frames taken by 8 cameras with more than 2 , 700 identities.”
Robust Person Re-Identification by Modelling Feature Uncertainty (2019, methodology): “The margin ranges from 1.98% mAP for DukeMTMC-ReID to 9.64% for CUHK01.”
Human Semantic Parsing for Person Re-identification (2018, methodology): “DukeMTMC-reID [50] dataset consists of the person images which are extracted from the DukeMTMC [31] tracking dataset.”
Cross-Domain Adversarial Feature Learning for Sketch Re-identification (2018, methodology): “Inspired by [29] and [24], we handled the dataset insufficient problem by pre-training on the public person Re-ID datasets, like Market1501 [36], CUHK03 [16] and Dukemtmc [22].”
FD-GAN: Pose-guided Feature Distilling GAN for Robust Person Re-identification (2018, methodology): “Market-1501 [5], CUHK03 [6] and DukeMTMC-reID [7] datasets.”
Rethinking the Competition Between Detection and ReID in Multiobject Tracking (2020, methodology): “ID F1 Score (IDF1) [40] is employed to evaluate the association performance.”
CityFlow: A City-Scale Benchmark for Multi-Target Multi-Camera Vehicle Tracking and Re-Identification (2019, methodology): “As for the evaluation of MTMC tracking, we adopted the metrics used by the MOTChallenge [5, 24] and DukeMTMC [34] benchmarks.”
CLIP-ReID: Exploiting Vision-Language Model for Image Re-Identification without Concrete Text Labels (2022, methodology): “We evaluate our method on four person re-identification datasets, including MSMT17 (Wei et al. 2018), Market-1501 (Zheng et al. 2015), DukeMTMC-reID (Ristani et al. 2016), Occluded-Duke (Miao et al. 2019), and two vehicle ReID datasets, VeRi-776 (Liu et al. 2016b) and VehicleID (Liu et al. 2016a).”
Joint Object Detection and Multi-Object Tracking with Graph Neural Networks (2021, methodology): “For MOT, we use standard CLEAR MOT metrics [68] and IDF1 metric [69].”
Learning Discriminative Features with Multiple Granularities for Person Re-Identification (2018, methodology): “DukeMTMC-reID This dataset is a subset of the DukeMTMC [27] used for person re-identification by images.”
Deep Domain-Adversarial Image Generation for Domain Generalisation (2020, methodology): “We use two commonly used re-ID datasets, namely Market1501 (Zheng et al. 2015) and DukeMTMC-reID (Duke) (Ristani et al. 2016; Zheng, Zheng, and Yang 2017).”
Image-Image Domain Adaptation with Preserved Self-Similarity and Domain-Dissimilarity for Person Re-identification (2017, methodology): “We select two large-scale re-ID datasets for experiment, i.e ., Market-1501 [49] and DukeMTMC-reID [36, 51].”