MOANA: An Online Learned Adaptive Appearance Model for Robust Multiple Object Tracking in 3D

Tang, Zheng; Hwang, Jenq‐Neng

doi:10.1109/access.2019.2903121

Cited by 36 publications

(15 citation statements)

References 52 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To demonstrate the tracking performance of our proposed system, we compare the performance with the following four different tracking methods: the MAST [44] and MOANA [45] are based on tracking by segmentation, and the other two methods RNN [46]and SORT [47] are based on tracking by detection. To fairly evaluate the tracking performance for each method, we adopt the following metrics which are widely used in Multiple Object Tracking(MOT) Challenge [48].…”

Section: A Multiple Vehicle Tracking Resultsmentioning

confidence: 99%

Deformable Model-Based Vehicle Tracking and Recognition Using 3-D Constrained Multiple-Kernels and Kalman Filter

Liu

2021

IEEE Access

View full text Add to dashboard Cite

Video-based vehicle tracking and recognition is an important application in the Intelligent Transportation System. High similarity among vehicle types, frequent occlusion and low video resolution in traffic surveillance are the major problems in this research area. In this paper, we proposed a vehicle tracking system by using 3-D constrained multiple-kernels, facilitated with Kalman filtering, to continuously update the location of the moving vehicles. To further robustly and efficiently track vehicles that are partially or even fully occluded, evolutionary optimization is applied to camera calibration for systematically building 3-D vehicle model, from which we can extract the vehicle's features such as the vehicle type, color and license plate. Then, a self-similarity descriptor is further introduced for vehicle re-identification. The proposed system is evaluated on the NVIDIA AI City Datasets and one self-recorded high-resolution video. The experimental results have shown the favorable performance, which not only can successfully track vehicles under occlusion, but also can maintain the knowledge of 3-D vehicle geometry.INDEX TERMS 3-D vehicle modeling, constrained multiple kernels, camera calibration, Kalman filter, vehicle tracking and recogniton,

show abstract

Section: A Multiple Vehicle Tracking Resultsmentioning

confidence: 99%

Deformable Model-Based Vehicle Tracking and Recognition Using 3-D Constrained Multiple-Kernels and Kalman Filter

Liu

2021

IEEE Access

View full text Add to dashboard Cite

show abstract

“…A convolutional neural network (CNN) has great advantages in representing visual data compared with traditional model-based and feature-based tracking algorithms, and they have been widely used in various computer vision tasks, such as image classification, semantic segmentation, and so on [31][32][33][34]. However, it is not easily applicable in visual tracking, since it is difficult to obtain useful datasets including diverse combination of targets and backgrounds with different appearance, and motion modes of different categories of targets in different video sequences.…”

Section: Moving-target-tracking Methods Based On Mdnetmentioning

confidence: 99%

Sea-Surface Target Visual Tracking with a Multi-Camera Cooperation Approach

Rao

Chen

et al. 2022

Sensors

View full text Add to dashboard Cite

Cameras are widely used in the detection and tracking of moving targets. Compared to target visual tracking using a single camera, cooperative tracking based on multiple cameras has advantages including wider visual field, higher tracking reliability, higher precision of target positioning and higher possibility of multiple-target visual tracking. With vast ocean and sea surfaces, it is a challenge using multiple cameras to work together to achieve specific target tracking and detection, and it will have a wide range of application prospects. According to the characteristics of sea-surface moving targets and visual images, this study proposed and designed a sea-surface moving-target visual detection and tracking system with a multi-camera cooperation approach. In the system, the technologies of moving target detection, tracking, and matching are studied, and the strategy to coordinate multi-camera cooperation is proposed. The comprehensive experiments of cooperative sea-surface moving-target visual tracking show that the method used in this study has improved performance compared with contrapositive methods, and the proposed system can meet the needs of multi-camera cooperative visual tracking of moving targets on the sea surface.

show abstract

“…Table VII shows the performance of our approach comparing with those of state-of-the-art SCT methods [45], [60], [61] in MTMCT. DeepSORT [60] is a Kalman-filter-based online tracking method.…”

Section: Sct Performancementioning

confidence: 99%

“…Tracklet Clustering (TC) [45] is an offline method, which is the winner of the AI City Challenge Workshop at CVPR 2018 [62]. MOANA [61] is the state-ofthe-art approach on the MOT Challenge 2015 3D benchmark. In CityFlow dataset, there are three available public detection results (i.e., SSD512 [63], YOLOv3 [64] and Faster R-CNN [65]).…”

Section: Sct Performancementioning

confidence: 99%

Multi-Target Multi-Camera Tracking of Vehicles Using Metadata-Aided Re-ID and Trajectory-Based Camera Link Model

Hsu

Cai

Wang

et al. 2021

IEEE Trans. on Image Process.

Self Cite

View full text Add to dashboard Cite

In this paper, we propose a novel framework for multi-target multi-camera tracking (MTMCT) of vehicles based on metadata-aided re-identification (MA-ReID) and the trajectory-based camera link model (TCLM). Given a video sequence and the corresponding frame-by-frame vehicle detections, we first address the isolated tracklets issue from single camera tracking (SCT) by the proposed traffic-aware singlecamera tracking (TSCT). Then, after automatically constructing the TCLM, we solve MTMCT by the MA-ReID. The TCLM is generated from camera topological configuration to obtain the spatial and temporal information to improve the performance of MTMCT by reducing the candidate search of ReID. We also use the temporal attention model to create more discriminative embeddings of trajectories from each camera to achieve robust distance measures for vehicle ReID. Moreover, we train a metadata classifier for MTMCT to obtain the metadata feature, which is concatenated with the temporal attention based embeddings. Finally, the TCLM and hierarchical clustering are jointly applied for global ID assignment. The proposed method is evaluated on the CityFlow dataset, achieving IDF1 76.77%, which outperforms the state-of-the-art MTMCT methods.

show abstract

MOANA: An Online Learned Adaptive Appearance Model for Robust Multiple Object Tracking in 3D

Cited by 36 publications

References 52 publications

Deformable Model-Based Vehicle Tracking and Recognition Using 3-D Constrained Multiple-Kernels and Kalman Filter

Deformable Model-Based Vehicle Tracking and Recognition Using 3-D Constrained Multiple-Kernels and Kalman Filter

Sea-Surface Target Visual Tracking with a Multi-Camera Cooperation Approach

Multi-Target Multi-Camera Tracking of Vehicles Using Metadata-Aided Re-ID and Trajectory-Based Camera Link Model

Contact Info

Product

Resources

About