Pedestrian Detection from Lidar Data via Cooperative Deep and Hand-Crafted Features

Lin, Tzu‐Chieh; Tan, Daniel Stanley; Tang, Hsueh-Ling; Chien, Shih‐Chieh; Chang, Feng-Chia; Chen, Yung-Yao; Cheng, Winton; Hua, Kai-Lung

doi:10.1109/icip.2018.8451578

Cited by 18 publications

(5 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Many studies have leveraged LiDAR data to tackle a broad range of challenges in the transportation engineering field. These challenges include assessing road conditions [ 10 , 11 , 12 , 13 , 14 ], monitoring traffic [ 15 , 16 , 17 ], aiding autonomous vehicle navigation [ 18 ], detecting pedestrians [ 19 , 20 , 21 , 22 ], facilitating transportation planning by mapping transportation network infrastructures [ 3 , 23 , 24 ], and other related applications.…”

Section: Literature Reviewmentioning

confidence: 99%

Leveraging Generative Design and Point Cloud Data to Improve Conformance to Passing Lane Layout

Momeni Rad,

Sydora,

El-Basyouny

2024

Sensors

View full text Add to dashboard Cite

Inadequate highway design is a leading cause of traffic accidents, underscoring the importance of adhering to guidelines and regulations for highway design. These standards exist to safeguard road users by addressing crucial factors, like road geometry, signage, and lane markings. Thus, emphasis is placed on computational methods that can optimize towards higher levels of safety, capacity, efficiency, and sustainability in highway designs. Building Information Modeling (BIM) enhances this process by creating a digital model with physical and operational attributes. In this study, a user-friendly, logic-based language is utilized to encode rules for designing highway passing lanes by which designs are automatically evaluated and generated in the BIM-kit software toolkit. This approach is applied to 16 real-world passing lanes in Alberta, showcasing its utility in transportation. The analysis reveals significant enhancements, with rule compliance increasing from 61.82% to 91.31% after employing generative design techniques. These findings underscore the significance of generative design in transportation, offering engineers an efficient tool to create innovative, compliant solutions for highway projects.

show abstract

Section: Literature Reviewmentioning

confidence: 99%

Leveraging Generative Design and Point Cloud Data to Improve Conformance to Passing Lane Layout

Momeni Rad,

Sydora,

El-Basyouny

2024

Sensors

View full text Add to dashboard Cite

show abstract

“…Pedestrian detection deals with the identification of pedestrians in the environment around an ego-vehicle. There exist approaches in the literature which perform pedestrian detection only using lidar sensors [438]. However, such approaches are usually not popular in the community due to fact that the features obtained from camera images are significantly richer as compared to the ones obtained from lidar or radar.…”

Section: Task Formulationmentioning

confidence: 99%

Knowledge Augmented Machine Learning with Applications in Autonomous Driving: A Survey

Wörmann¹,

Bogdoll²,

Bührle³

et al. 2022

Preprint

View full text Add to dashboard Cite

The existence of representative datasets is a prerequisite of many successful artificial intelligence and machine learning models. However, the subsequent application of these models often involves scenarios that are inadequately represented in the data used for training. The reasons for this are manifold and range from time and cost constraints to ethical considerations. As a consequence, the reliable use of these models, especially in safety-critical applications, is a huge challenge. Leveraging additional, already existing sources of knowledge is key to overcome the limitations of purely data-driven approaches, and eventually to increase the generalization capability of these models. Furthermore, predictions that conform with knowledge are crucial for making trustworthy and safe decisions even in underrepresented scenarios. This work provides an overview of existing techniques and methods in the literature that combine data-based models with existing knowledge. The identified approaches are structured according to the categories integration, extraction and conformity. Special attention is given to applications in the field of autonomous driving.Acknowledgement: The research leading to these results is funded by the German Federal Ministry for Economic Affairs and Climate Action within the project "KI Wissen -Entwicklung von Methoden für die Einbindung von Wissen in maschinelles Lernen". The authors would like to thank the consortium for the successful cooperation.

show abstract

“…Tang et al [11] and Lin et al [12] approaches are sim- We present a pedestrian detector in 3D point clouds based on PointNet++, a hierarchical neural network that extracts deep features from the 3D geometric information of a point cloud. The local features are extracted capturing fine geometric structures from small neighborhoods.…”

Section: Related Workmentioning

confidence: 99%

“…Tang et al [11] and Lin et al [12] present their results using the KITTI dataset. Although Beamagine datasets present different properties (for example in terms of vertical resolution), a comparison between these methods and the one proposed in our paper (Table 5) gives a rough view of how does our approach detecting pedestrians using Point-Net++ compare with them.…”

Section: Comparison With State-of-the-art Approaches On Different Dat...mentioning

confidence: 99%

Pedestrian Detection in 3D Point Clouds using Deep Neural Networks

Lorente,

Casas,

Royo

et al. 2021

Preprint

View full text Add to dashboard Cite

Detecting pedestrians is a crucial task in autonomous driving systems to ensure the safety of drivers and pedestrians. The technologies involved in these algorithms must be precise, robust and reliable, regardless of environment conditions. Relying solely on RGB cameras may not be enough to recognize road environments in situations where cameras cannot capture scenes properly. Some approaches aim to compensate for these limitations by combining RGB cameras with Time Of Flight (TOF) sensors, such as Light Detection and Ranging (LIDAR), which map the scene into 3D point clouds using pulses of light. However, there are few works that address this problem using exclusively the 3D geometric information provided by LIDARs. In this paper, we propose a PointNet++ based architecture to detect pedestrians in dense 3D point clouds. The aim is to explore the potential contribution of geometric information alone in pedestrian detection systems. PointNet++ is a deep neural network that learns global and local features of a point cloud from its geometry. This allows discerning the different patterns that characterize each class. In our case, the network must learn to classify pedestrians, so we need to carry out supervised training using accurately labeled pedestrian and non-pedestrian point cloud clusters as input. Therefore, we also present a semi-automatic labeling system to generate a point cloud dataset with ground truth. The labeling system transfers pedestrian and nonpedestrian labels from RGB images onto the 3D domain. The fact that our datasets have RGB registered with point clouds enables label transferring by back projection from 2D bounding boxes to point clouds, with only a light manual supervision to validate results. We train PointNet++ with the geometry of the resulting 3D labelled clusters. The evaluation confirms the effectiveness of the proposed method, yielding precision and recall values around 98%. This corroborates the hypothesis stated in the paper that 3D geometric information is essential for a neural network to learn to detect pedestrians in outdoor scenes, thus proving that LIDAR sensors should be a must in these systems.

show abstract

Pedestrian Detection from Lidar Data via Cooperative Deep and Hand-Crafted Features

Cited by 18 publications

References 11 publications

Leveraging Generative Design and Point Cloud Data to Improve Conformance to Passing Lane Layout

Leveraging Generative Design and Point Cloud Data to Improve Conformance to Passing Lane Layout

Knowledge Augmented Machine Learning with Applications in Autonomous Driving: A Survey

Pedestrian Detection in 3D Point Clouds using Deep Neural Networks

Contact Info

Product

Resources

About