HeadNet: Pedestrian Head Detection Utilizing Body in Context

Chen, Gang; Cai, Xufen; Han, Hu; Shan, Shiguang; Chen, Xilin

doi:10.1109/fg.2018.00089

Cited by 9 publications

(15 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Zhang et al [26] made great efforts on analyzing the failures for a top-performing detector and provided how to engineer better detectors. In addition, HeadNet [21] aimed to utilize the body in context information, i.e., the relationship between individual body parts. Body part detection can be very useful in the succeeding human analysis tasks, such as action recognition.…”

Section: Pedestrian Detectionmentioning

confidence: 99%

“…In particular, the relation learning model in [17] considers the relations not only between objects but also between the whole image and individual objects. HeadNet [21] utilized the spatial semantics of the entire body as contextual information. These methods can utilize the relationship between individual objects effectively, but not explicitly model the semantic relationship between individual body parts.…”

Section: Object Relation Learningmentioning

confidence: 99%

“…The entire BPIF is followed by two convolutional layers for feature dimensionality reduction. It is worth noting that unlike HeadNet [21] the Body-in-Context (BiC) and Refinement modules in our model do not share parameters. Our considerations are as follows.…”

Section: Body Part Indexed Feature (Bpif)mentioning

confidence: 99%

“…We use the CUHK-SYSU Person Search Dataset [25] for experimental evaluations, which contains 18,184 images and 99,809 annotated pedestrian bounding boxes. The original dataset does not include body parts annotations, but Chen et al [21] provided head, head-shoulder, and upper body bounding boxes annotations for 16,907 images from CUHK-SYSU Person Search Dataset. Following the protocol in [21], we use 13,399 and 3508 images as the training and testing sets, respectively.…”

Section: Datasets and Settingsmentioning

confidence: 99%

“…Ref. [21] proposes a HeadNet to jointly detect pedestrian and the head, head-shoulder, and upper body in a multi-task way [22,23] by utilizing the body context information. But the performance of joint pedestrian and body part detection in the case of severe overlap/occlusion, remains unsatisfactory.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Joint Pedestrian and Body Part Detection via Semantic Relationship Learning

Lan

Chen

et al. 2019

Applied Sciences

Self Cite

View full text Add to dashboard Cite

While remarkable progress has been made to pedestrian detection in recent years, robust pedestrian detection in the wild e.g., under surveillance scenarios with occlusions, remains a challenging problem. In this paper, we present a novel approach for joint pedestrian and body part detection via semantic relationship learning under unconstrained scenarios. Specifically, we propose a Body Part Indexed Feature (BPIF) representation to encode the semantic relationship between individual body parts (i.e., head, head-shoulder, upper body, and whole body) and highlight per body part features, providing robustness against partial occlusions to the whole body. We also propose an Adaptive Joint Non-Maximum Suppression (AJ-NMS) to replace the original NMS algorithm widely used in object detection, leading to higher precision and recall for detecting overlapped pedestrians. Experimental results on the public-domain CUHK-SYSU Person Search Dataset show that the proposed approach outperforms the state-of-the-art methods for joint pedestrian and body part detection in the wild.

show abstract

Section: Pedestrian Detectionmentioning

confidence: 99%

Section: Object Relation Learningmentioning

confidence: 99%

Section: Body Part Indexed Feature (Bpif)mentioning

confidence: 99%

Section: Datasets and Settingsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Joint Pedestrian and Body Part Detection via Semantic Relationship Learning

Lan

Chen

et al. 2019

Applied Sciences

Self Cite

View full text Add to dashboard Cite

show abstract

Feature Enhancement for Joint Human and Head Detection

Zhang

Zhuang

et al. 2019

Biometric Recognition

View full text Add to dashboard Cite

Human and head detection have been rapidly improved with the development of deep convolutional neural networks. However, these two detection tasks are often studied separately, without taking advantage of the relationship between human and head. In this paper, we present a new two-stage detection framework, namely Joint Enhancement Detection (JED), to simultaneously detect human and head based on enhanced features. Specifically, the proposed JED contains two newly added modules, i.e., the Body Enhancement Module (BEM) and the Head Enhancement Module (HEM). The former is designed to enhance the features used for human detection, while the latter aims to enhance the features used for head detection. With these enhanced features in a joint framework, the proposed method is able to detect human and head simultaneously and efficiently. We verify the effectiveness of the proposed method on the CrowdHuman dataset and achieve better performance than baseline method for both human and head detection.

show abstract

HAPNet: a head-aware pedestrian detection network associated with the affinity field

Ding

Zhao

et al. 2022

Sci. China Inf. Sci.

View full text Add to dashboard Cite

HeadNet: Pedestrian Head Detection Utilizing Body in Context

Cited by 9 publications

References 18 publications

Joint Pedestrian and Body Part Detection via Semantic Relationship Learning

Joint Pedestrian and Body Part Detection via Semantic Relationship Learning

Feature Enhancement for Joint Human and Head Detection

HAPNet: a head-aware pedestrian detection network associated with the affinity field

Contact Info

Product

Resources

About