A New Literature Review of 3D Object Detection on Autonomous Driving
Authors: Peng Zhang, Xin Li, Xin Lin, Liang He
JAIR 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | This review delivers an extensive overview of the emerging technologies in 3D object detection tailored for autonomous vehicles. It encompasses a thorough examination, evaluation, and integration of the current research landscape in this domain, staying up-to-date with the latest advancements in 3D object detection and suggesting prospective avenues for future research. Our survey begins by clarifying the principles of 3D object detection and addressing its present challenges in the 3D domain. We then introduce three distinct taxonomies: camera-based, point cloud-based, and multi-modality-based approaches, providing a comprehensive classification of contemporary 3D object detection methodologies from various angles. Diverging from previous reviews, this paper also highlights and scrutinizes common issues and solutions for specific scenarios (such as pedestrian detection, lane lines, roadside cameras, and weather conditions) in object detection. Furthermore, we conduct an in-depth analysis and comparison of different classifications and methods, utilizing various datasets and experimental outcomes. Conclusively, we suggest several potential research directions, offering valuable insights for the ongoing evolution of 3D object detection technology. This review aims to serve as a comprehensive resource for researchers and practitioners in the field, guiding future innovations in 3D object detection for autonomous driving. |
| Researcher Affiliation | Academia | Peng Zhang EMAIL Xin Li (Corresponding author) EMAIL Xin Lin (Corresponding author) EMAIL Liang He EMAIL East China Normal University, China |
| Pseudocode | No | The paper is a literature review and does not present new algorithms in pseudocode format. It describes various methods conceptually but does not include any structured pseudocode or algorithm blocks. |
| Open Source Code | No | This paper is a literature review that synthesizes existing research and does not present a new methodology for which source code would be released. Therefore, there is no statement about releasing code or a link to a repository for the work described in this paper. |
| Open Datasets | Yes | The leading 3D object detection datasets relevant to autonomous driving currently include KITTI (Geiger et al., 2013), nu Scenes (Caesar et al., 2020), Waymo Open (Sun et al., 2020), Argoverse (Chang et al., 2019), Apollo Scape (Huang et al., 2019), Panda Set (Xiao et al., 2021), H3D (Patil et al., 2019), and more. We have compiled the most recent datasets for 3D object detection, as detailed in Table 1. |
| Dataset Splits | Yes | nu Scenes comprises 700 training scenes, 150 validation scenes, and 150 testing scenes. In addition to this, it provides annotations for object-level attributes such as visibility, activity, pose, and more. It includes a large volume of RGB and pointcloud data (in PCD format). Given the substantial size of the complete nu Scenes dataset, users often prefer the nu Scenes-mini dataset. Waymo Open Dataset Waymo includes a total of 798 training scenes and 202 validation scenes, each with 2D and 3D annotated labels. |
| Hardware Specification | No | This paper is a literature review that analyzes and compares existing research. It does not conduct its own experiments or present new methods that would require specific hardware for implementation or evaluation. Therefore, no hardware specifications are provided. |
| Software Dependencies | No | This paper is a literature review and does not present new experimental results requiring specific software implementations. Therefore, it does not list any software dependencies or versions for its own work. |
| Experiment Setup | No | This paper is a literature review and does not conduct its own experiments. Therefore, no specific experimental setup details, hyperparameters, or training configurations are provided for any new methodology. |