A real-time Deep Learning pedestrian detector for robot navigation

Ribeiro, David Augusto; Mateus, André; Miraldo, Pedro; Nascimento, Jacinto C.

doi:10.1109/icarsc.2017.7964070

Cited by 31 publications

(15 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This allows us to know the location of the pedestrian in the 3D environment, i.e., the pedestrian's position relative to the camera system can be obtained. This is a useful feature in human-awareness navigation [8], [44].…”

Section: B Contributionsmentioning

confidence: 99%

“…Along with the availability of larger image datasets (e.g., [1], [2], [3], [4]) and the increase of hardware capabilities, based on GPUs, we have observed a significant improvement in detection and classification [5]. Nowadays, pedestrian analysis is mainly addressed using Deep Learning (DL) techniques (e.g., [6], [7], [8]) which attempt to learn high level abstractions of the data. One of the most important and well-known techniques in DL are the Convolutional Neural Networks (CNNs) [6], which halved the error rate for image classification.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

OmniDRL: Robust Pedestrian Detection using Deep Reinforcement Learning on Omnidirectional Cameras

Pais

Dias

Nascimento

et al. 2019

2019 International Conference on Robotics and Automation (ICRA)

Self Cite

View full text Add to dashboard Cite

Pedestrian detection is one of the most explored topics in computer vision and robotics. The use of deep learning methods allowed the development of new and highly competitive algorithms. Deep Reinforcement Learning has proved to be within the state-of-the-art in terms of both detection in perspective cameras and robotics applications. However, for detection in omnidirectional cameras, the literature is still scarce, mostly because of their high levels of distortion. This paper presents a novel and efficient technique for robust pedestrian detection in omnidirectional images. The proposed method uses deep Reinforcement Learning that takes advantage of the distortion in the image. By considering the 3D bounding boxes and their distorted projections into the image, our method is able to provide the pedestrian's position in the world, in contrast to the image positions provided by most state-of-the-art methods for perspective cameras. Our method avoids the need of preprocessing steps to remove the distortion, which is computationally expensive. Beyond the novel solution, our method compares favorably with the state-of-the-art methodologies that do not consider the underlying distortion for the detection task.1 Although undistorted images keep the perspective projection constraints (i.e., straight lines in the world will be projected into straight lines in the image), the objects will be stretched. This means that the objects should not be approximated by regular bounding boxes, which are used in most of the DL techniques for object detection (we note that there are alternatives that do not consider regular bounding boxes in [19]).

show abstract

Section: B Contributionsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

OmniDRL: Robust Pedestrian Detection using Deep Reinforcement Learning on Omnidirectional Cameras

Pais

Dias

Nascimento

et al. 2019

2019 International Conference on Robotics and Automation (ICRA)

Self Cite

View full text Add to dashboard Cite

show abstract

“…In Ribeiro et al, 26 for example, differently from our work, the use of GPU is mandatory to perform human detection and to accomplish the demands of their application. Besides, their human detection method is not delivered as a service that can be flexibly used by different applications.…”

Section: Human Detectionmentioning

confidence: 99%

“…Besides, their human detection method is not delivered as a service that can be flexibly used by different applications. The solution proposed in Ribeiro et al 26 is not designed to be distributed over the nodes of the infrastructure; thus, it is not concerned with issues such as synchronism, scalability and reliability. They use only one powerful node (GPU) to have a low false positive rate, while we use simple techniques and the redundant information provided by a network of cameras to achieve this goal.…”

Section: Human Detectionmentioning

confidence: 99%

“…It is important to mention that the focus of the present work is not to propose a generic state of the art human detection system as Zhang et al 25 We also do not aim to provide just human position information for autonomous robots as can be seen in Ribeiro et al 26 or Lee et al 21 But as mentioned before, our human detection service provides two different features: (1) a service with properties as flexibility, scalability, parallelism, and reliability and also (2) human location, given by a multi-camera network, that allows us to avoid some problems that make single camera human detectors fail in real world applications. The interrelationship of the human detection service and the camera network is handled to meet the time requirements of the applications and real-time execution is achieved.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A flexible human detection service suitable for Intelligent Spaces based on a multi-camera network

Almonfrey

Carmo

Queiroz

et al. 2018

International Journal of Distributed Sensor Networks

View full text Add to dashboard Cite

The research field of the Intelligent Spaces has experienced increasing attention in the last decade. As an instance of the ubiquitous computing paradigm, the general idea is to extract information from the ambient and use it to interact and provide services to the actors present in the environment. The sensory analysis is mandatory in this area and humans are usually the principal actors involved. In this sense, we propose a human detector to be used in an Intelligent Space based on a multi-camera network. Our human detector is implemented in the same paradigm of our Intelligent Space. As a contribution of the present work, the human detector is designed to be a service that is scalable, reliable and parallelizable. It is also a concern of our service to be flexible, less structured as possible, attending different Intelligent Space applications and services, as well as their requirements. As it can be found in different everyday environments, a multicamera system is used to overcome some difficulties traditionally faced by existing human detection approaches. To validate our approach, we implement three different applications that are proof of concept of many day-to-day real tasks. Two of these applications involve human-robot interaction. With respect to time and detection performance requirements, our human detection service has proved to be suitable for interacting with the other services of our Intelligent Space, in order to successfully complete the tasks of each application.

show abstract

Pedestrian Detection Using ACF Based Fast R-CNN

Zhuang

2018

Communications in Computer and Information Science

View full text Add to dashboard Cite

A real-time Deep Learning pedestrian detector for robot navigation

Cited by 31 publications

References 16 publications

OmniDRL: Robust Pedestrian Detection using Deep Reinforcement Learning on Omnidirectional Cameras

OmniDRL: Robust Pedestrian Detection using Deep Reinforcement Learning on Omnidirectional Cameras

A flexible human detection service suitable for Intelligent Spaces based on a multi-camera network

Pedestrian Detection Using ACF Based Fast R-CNN

Contact Info

Product

Resources

About