2019
DOI: 10.48550/arxiv.1906.10313
|View full text |Cite
Preprint
|
Sign up to set email alerts
|

DensePeds: Pedestrian Tracking in Dense Crowds Using Front-RVO and Sparse Features

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

0
14
0

Year Published

2020
2020
2020
2020

Publication Types

Select...
6

Relationship

6
0

Authors

Journals

citations
Cited by 7 publications
(14 citation statements)
references
References 0 publications
0
14
0
Order By: Relevance
“…Pedestrian tracking has been widely applied in surveillance video analysis and is well developed based on research on multi-object tracking problems [4]- [7].…”
Section: A Perception For Surveillance Systemsmentioning
confidence: 99%
“…Pedestrian tracking has been widely applied in surveillance video analysis and is well developed based on research on multi-object tracking problems [4]- [7].…”
Section: A Perception For Surveillance Systemsmentioning
confidence: 99%
“…The detection bounding boxes are used as inputs to DensePeds [26], a state-of-the-art pedestrian tracking algorithm that assigns a unique ID to each detected pedestrian across multiple consecutive images. This ID is used to compute each pedestrian's position in the image over time.…”
Section: Pedestrian Detection and Trackingmentioning
confidence: 99%
“…In the deep learning era, interesting methods like [39], [47], [31], [35] have been proposed to this end. Traffic trajectory prediction [9,13,14] and tracking [8,[10][11][12], have been widely studied robotics and computer vision, whereas driver behavior modeling has mostly been restricted to traffic psychology and the social sciences [1-4, 7, 15-17, 19, 20, 22, 25, 33, 34, 34, 38, 42, 45, 48].…”
Section: Driving Scene Understandingmentioning
confidence: 99%
“…We stack 3 layers of ST-GCN i.e. the pair of equations (8) and (9) with temporal span τ = 3. The last layer of ST-GCN is followed by the classifier that includes average pooling and fully connected layers to obtain the classification logits, which are further converted into class probabilities using the softmax layer.…”
Section: Implementation Detailsmentioning
confidence: 99%