UA-DETRAC: A new benchmark and protocol for multi-object detection and tracking

By Longyin Wen, Dawei Du, Zhaowei Cai, Zhen Lei, Ming-Ching Chang, Honggang Qi, Jongwoo Lim, Ming-Hsuan Yang, Siwei Lyu. 2015

749 citations recorded by the source corpus.

In recent years, numerous effective multi-object tracking (MOT) methods are developed because of the wide range of applications. Existing performance evaluations of MOT methods usually separate the object tracking step from the object detection step by using the same fixed object detection results for comparisons. In this work, we perform a comprehensive quantitative study on the effects of object detection accuracy to the overall MOT performance, using the new large-scale University at Albany DETection and tRACking (UA-DETRAC) benchmark dataset. The UA-DETRAC benchmark dataset consists of 100 challenging video sequences captured from real-world traffic scenes (over 140,000 frames with rich annotations, including occlusion, weather, vehicle category, truncation, and vehicle bounding boxes) for object detection, object tracking and MOT system. We evaluate complete MOT systems constructed from combinations of state-of-the-art object detection and object tracking methods. Our analysis shows the complex effects of object detection accuracy on MOT system performance. Based on these observations, we propose new evaluation tools and metrics for MOT systems that consider both object detection and object tracking for comprehensive analysis.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Evaluation & Benchmarks

How researchers see this paper

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

In Artificial Intelligence

In Computer Vision and Pattern Recognition

In Evaluation & Benchmarks

Citations in the corpus

Benchmarking Robustness in Object Detection: Autonomous Driving when Winter is Coming (2019, background): “…conditions [Sakaridis et al., 2018a,b, Volk et al., 2019, Bernuth et al., 2019, Hospach et al., 2016, Bernuth et al., 2018] or collecting real [Wen et al., 2015, Yu et al., 2018, Che et al., 2019, Caesar et al., 2019] and artiﬁcial [Gaidon et al., 2016, Ros et al., 2016, Richter et al.,…”
Joint Monocular 3D Vehicle Detection and Tracking (2018, background): “KITTI-D [12] D C,P 7k - 41k X X KAIST [20] D P 50k - 42k X X KITTI-T [12] T C 8k 938 65k X X X MOT15-3D [24] T P 974 29 5632 X X MOT15-2D [24] T P 6k 500 40k X MOT16 [29] T C,P 5k 467 110k X X UA-DETRAC [45] D,T C 84k 6k 578k X X”
Deep learning in video multi-object tracking: A survey (2019, methodology): “The method reached top performance on the UA-DETRAC dataset [32], but the performance on MOT16 was not very good when compared with other algorithms using private detections.”
HOTA: A Higher Order Metric for Evaluating Multi-object Tracking (2020, methodology): “Many of these rank trackers using theMOTAmetric (PANDA [77], BDD100k [87], Waymo [73], ArgoVerse [14], PoseTrack [1], MOTS [74]), or a variation of MOTA (nuScenes [12], UA-DETRAC [78]).”
STREETS: A Novel Camera Network Dataset for Traffic Flow (2019, background): “Popular examples such as KITTI [8], MOTChallenge 2015 [9], and UA-DETRAC [10] present benchmark computer vision tasks that are intended for autonomous vehicles and object tracking.”
3D Multi-Object Tracking: A Baseline and New Evaluation Metrics (2019, neutral): “proposed in the independent work [37].”
Opening up Open World Tracking (2021, background): “Existing MOT datasets focus on closed-set tracking on video [18, 24, 83, 90] or LiDAR streams [14, 75].”
MOTS: Multi-Object Tracking and Segmentation (2019, background): “UA-DETRAC [58, 36] also features street scenes but contains annotations for vehicles only.”
Recent Advancements in End-to-End Autonomous Driving Using Deep Learning: A Survey (2023, neutral): “UA-DETRAC [126] 2015 ✓ ✓ ✓ 140k frames Beijing, Tianjing CC BY-NC-SA 3.”
Drone-Based RGB-Infrared Cross-Modality Vehicle Detection Via Uncertainty-Aware Learning (2020, background): “DOTA [8], UA-DETRAC [20], and NightSurveillance [21] are representative object detection datasets for the RGB modality.”
Deep Affinity Network for Multiple Object Tracking (2018, methodology): “ay appear less challenging than MOT datasets, the larger size of data and the scenes of crossings and junctions make tracking in this dataset a difﬁcult task. TABLE 8: Attributes of UA-DETRAC dataset [30], [31]. The ‘Boxes’ column indicates the total number of bounding boxes in videos. ‘Length’ is g”
Vision Meets Drones: A Challenge (2018, methodology): “Among many factors and efforts that lead to the fast evolution of computer vision techniques, a notable contribution should be attributed to the invention or organization of numerous benchmarks, such as Caltech [1], KITTI [2], ImageNet [3], and MS COCO [4] for object detection, and OTB [5], VOT [6],”
EDF: Ensemble, Distill, and Fuse for Easy Video Labeling (2018, methodology): “Datasets: We evaluate our system on two datasets: KITTI [5] and DETRAC [29].”