Recognizing pedestrian’s unsafe behaviors in far-infrared imagery at night

Lee, Eun Ju; Ko, Byoung Chul; Nam, Jae-Yeal

doi:10.1016/j.infrared.2016.03.006

Cited by 37 publications

(21 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Their CNN is a modification of GoogLeNet but has a more compact structure. Lee et al [20] designed a lightweight CNN consisting of two convolutional layers and two subsampling layers for recognizing unsafe behaviors of pedestrians using thermal images captured from moving vehicles at night. They combined their lightweight CNN with a boosted random forest classifier.…”

Section: Related Workmentioning

confidence: 99%

Borrow From Anywhere: Pseudo Multi-Modal Object Detection in Thermal Imagery

Devaguptapu

Akolekar

Sharma

et al. 2019

2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)

118

View full text Add to dashboard Cite

Can we improve detection in the thermal domain by borrowing features from rich domains like visual RGB? In this paper, we propose a 'pseudo-multimodal' object detector trained on natural image domain data to help improve the performance of object detection in thermal images. We assume access to a large-scale dataset in the visual RGB domain and relatively smaller dataset (in terms of instances) in the thermal domain, as is common today. We propose the use of well-known image-to-image translation frameworks to generate pseudo-RGB equivalents of a given thermal image and then use a multi-modal architecture for object detection in the thermal image. We show that our framework outperforms existing benchmarks without the explicit need for paired training examples from the two domains. We also show that our framework has the ability to learn with less data from thermal domain when using our approach.

show abstract

Section: Related Workmentioning

confidence: 99%

Borrow From Anywhere: Pseudo Multi-Modal Object Detection in Thermal Imagery

Devaguptapu

Akolekar

Sharma

et al. 2019

2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)

118

View full text Add to dashboard Cite

show abstract

“…Object detection and classification in thermal imagery has been an active area of research for a number of years. [25][26][27][28][29] A range of trial work in the literature addresses the task of detecting people and objects in thermal imagery. 28,29 An early method such as a probabilistic template-based approach 28 is proposed for pedestrian detection by using far-infrared (thermal) images.…”

Section: Related Workmentioning

confidence: 99%

Visible to infrared transfer learning as a paradigm for accessible real-time object detection and classification in infrared imagery

Gaus

Bhowmik

Isaac-Medina

et al. 2020

Counterterrorism, Crime Fighting, Forensics, and Surveillance Technologies IV

View full text Add to dashboard Cite

show abstract

“…Among all road users, pedestrians increase the risk of road accidents because they are so exible and unpredictable (5,6). In many studies, several behavioral factors like talking on the cellphone while crossing the street, neglecting tra c light, crossing a red light and crossing at inappropriate places (crossing dangerous lines) are suggested as pedestrians' unsafe behaviors (7)(8)(9)(10). Furthermore, low perceived risk of unsafe behaviors (11,12), consumption of unauthorized drinks, poor visibility on the roads (13,14) and distraction (9) are also reported as other behavior-associated factors that endanger pedestrian safety.…”

Section: Introductionmentioning

confidence: 99%

What are the challenges of pedestrian safety from the viewpoints of traffic and transport stakeholders? A qualitative Study

Haghighi¹,

Aghdam²,

Sadeghi‐Bazargani³

et al. 2020

Preprint

View full text Add to dashboard Cite

Background Pedestrians are among the most vulnerable groups in traffic accidents. Our aim in this study was to explore the challenges associated to pedestrian safety from the perspective of traffic and transport stakeholders. Methods In 2018, applying a qualitative approach, twenty-four traffic and transport stakeholders were invited to participate in semi-structured individual interviews in Tabriz, Iran. To analyze data, conventional content analysis approach was used. MAXQDA software version 11 was applied to manage data analysis process. Findings: Participants reported a wide range of challenges which were grouped into six categories: "Challenges related to pedestrians", "Challenges related to drivers", "Management system challenges", "Environmental infrastructure challenges", "Educational and media challenges", and "Challenges of legislation and enforcement". Conclusion We identified pedestrian safety as a challenging urban traffic and transport issue with specific complexities, particularly in the management system. With a holistic approach to the challenges, as discussed inside, all reported obstacles seem to be overshadowed by one core challenge, namely the lack of a traffic management system with health-oriented approach and enough authority. Using evidence while policy-making and intervention planning, as well as media support is recommended.

show abstract

Recognizing pedestrian’s unsafe behaviors in far-infrared imagery at night

Cited by 37 publications

References 23 publications

Borrow From Anywhere: Pseudo Multi-Modal Object Detection in Thermal Imagery

Borrow From Anywhere: Pseudo Multi-Modal Object Detection in Thermal Imagery

Visible to infrared transfer learning as a paradigm for accessible real-time object detection and classification in infrared imagery

What are the challenges of pedestrian safety from the viewpoints of traffic and transport stakeholders? A qualitative Study

Contact Info

Product

Resources

About