Li3DeTr: A LiDAR based 3D Detection Transformer

Erabati, Gopi Krishna; Araújo, Hélder

doi:10.1109/wacv56688.2023.00423

Cited by 14 publications

(4 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This technology enhances robot autonomy and self-driving cars. Studies include pointnet++ [44], voteNET [45], and 3DETR [46,47] for 3D object detection. Real-time 3D object detection, exemplified by Media pipe objectron [48,49], is an active area of research.…”

Section: D Object Detectionsmentioning

confidence: 99%

Enhancing Robot Inclusivity in the Built Environment: A Digital Twin-Assisted Assessment of Design Guideline Compliance

Ezhilarasu,

Pey,

Muthugala

et al. 2024

Buildings

View full text Add to dashboard Cite

Developing guidelines for designing robot-inclusive spaces has been challenging and resource-intensive, primarily relying on physical experiments and observations of robot interactions within the built environment. These conventional methods are often costly, time-consuming, and labour-intensive, demanding manual intervention. To address these limitations, this study explores the potential of using digital twins as a promising solution to offer detailed insights, reducing the dependence on physical experiments for studying robot-built environment interactions.Although the concept of digital twins is popular in many domains, the use of digital twins for this specific problem has not been explored yet. A novel methodology for assessing existing built environment guidelines by incorporating them as an architectural digital twin asset within robot simulation software is proposed in this regard. By analysing the digital interactions between robots and the architectural digital twin assets in simulations, the compatibility of the environment with robots is evaluated, ultimately contributing to enhancing these guidelines to be robot-inclusive. The ultimate goal is to create environments that are not only inclusive but also readily accessible to Autonomous Mobile Robots (AMRs). With this objective, the proposed methodology is tested on robots of different specifications to understand the robots’ interactions with different architectural digital twin environments and obstacles. The digital twin effectively demonstrates the capability of the proposed approach in assessing the robots’ suitability for deployment in the simulated environments. The gained insights contribute to improved comprehension and strengthen the existing design guidelines.

show abstract

Section: D Object Detectionsmentioning

confidence: 99%

Enhancing Robot Inclusivity in the Built Environment: A Digital Twin-Assisted Assessment of Design Guideline Compliance

Ezhilarasu,

Pey,

Muthugala

et al. 2024

Buildings

View full text Add to dashboard Cite

show abstract

“…Additionally, Refs. [37,38] apply the Detection Transformer (DETR) to LiDAR 3D detection tasks, generating predictions for each object query.…”

Section: Three-dimensional Object Detectionmentioning

confidence: 99%

Ship Detection in Maritime Scenes under Adverse Weather Conditions

Zhang,

Wang,

Meng

et al. 2024

Remote Sensing

View full text Add to dashboard Cite

Point cloud-based detection focuses on land traffic, rarely marine, facing issues with ships: it struggles in bad weather due to reliance on adverse weather data and fails to detect ships effectively due to overlooking size and appearance differences. Addressing the above challenges, our work introduces point cloud data of marine scenarios under realistically simulated adverse weather conditions and a dedicated Ship Detector tailored for marine environments. To adapt to various maritime weather conditions, we simulate realistic rain and fog in collected marine scene point cloud data. Additionally, addressing the issue of losing geometric and height information during feature extraction for large objects, we propose a Ship Detector. It employs a dual-branch sparse convolution layer for extracting multi-scale 3D feature maps, effectively minimizing height information loss. Additionally, a multi-scale 2D convolution module is utilized, which encodes and decodes feature maps and directly employs 3D feature maps for target prediction. To reduce dependency on existing data and enhance model robustness, our training dataset includes simulated point cloud data representing adverse weather conditions. In maritime point cloud ship detection, our Ship Detector, compared to adjusted small object detectors, demonstrates the best performance.

show abstract

“…Box-Attention enables the transformer model to better learn spatial information, thereby improving its performance in various visual tasks. Li3DeTr [27] can directly predict the category, position, and size of 3D objects from the LiDAR point cloud data. This work is the first to apply the deformable attention architecture to LiDAR-based 3D object detection models.…”

Section: Egocentric Perceptionmentioning

confidence: 99%

CoFormerNet: A Transformer-Based Fusion Approach for Enhanced Vehicle-Infrastructure Cooperative Perception

Li,

Zhao,

Tan

2024

Sensors

View full text Add to dashboard Cite

Vehicle–infrastructure cooperative perception is becoming increasingly crucial for autonomous driving systems and involves leveraging infrastructure’s broader spatial perspective and computational resources. This paper introduces CoFormerNet, which is a novel framework for improving cooperative perception. CoFormerNet employs a consistent structure for both vehicle and infrastructure branches, integrating the temporal aggregation module and spatial-modulated cross-attention to fuse intermediate features at two distinct stages. This design effectively handles communication delays and spatial misalignment. Experimental results using the DAIR-V2X and V2XSet datasets demonstrated that CoFormerNet significantly outperformed the existing methods, achieving state-of-the-art performance in 3D object detection.

show abstract

Li3DeTr: A LiDAR based 3D Detection Transformer

Cited by 14 publications

References 45 publications

Enhancing Robot Inclusivity in the Built Environment: A Digital Twin-Assisted Assessment of Design Guideline Compliance

Enhancing Robot Inclusivity in the Built Environment: A Digital Twin-Assisted Assessment of Design Guideline Compliance

Ship Detection in Maritime Scenes under Adverse Weather Conditions

CoFormerNet: A Transformer-Based Fusion Approach for Enhanced Vehicle-Infrastructure Cooperative Perception

Contact Info

Product

Resources

About