Thereafter, the identified literature is discussed and analyzed in detail. 2, we provide all relevant foundations and subsequently refer to related work in Sect. Throughout our review, we describe representative examples along the 3DOD pipeline, while highlighting seminal achievements. In total, our literature corpus comprises more than hundred papers which we examined in detail to provide a classification of all approaches. To carry out our review, we investigated papers that were published in a period from 2012 to 2021. Guo et al., Fernandes et al., ), which often focus on a particular domain (e.g., autonomous driving), specific data input (e.g., point cloud data), or a certain set of methods (e.g., deep learning techniques). With our work, we complement related surveys in the field (e.g., Arnold et al. For this reason, we present a comprehensive review of 3DOD models and methods with exemplary applications and aim to conceptualize the full range of 3DOD approaches along a multi-stage pipeline. Providing an overview about relevant approaches and seminal achievements may offer orientation and can help to initiate further development in the research community. Since 3DOD is a steadily growing field of investigation, there are several promising approaches and trends, including a large pool of various design options for the object detection pipeline. Against the backdrop of highly sophisticated 2DOD models, it is apparent that the focus of research is shifting to 3DOD as the necessary hardware in terms of sensors and computing units becomes increasingly available. The literature volume for 3DOD has increased significantly over the past years. 3DOD now extends this approach into the three-dimensional space by adding the desired parameters of dimension and orientation of the object to the established location and classification results. However, to fully grasp the scene in a real 3D world, 2D recognition and detection results alone are no longer sufficient. Meanwhile, 2D object detection (2DOD) has obtained impressive results in terms of precision and inference time, and is able to compete with or even surpass human vision. This objective is well known as 3D object detection (3DOD). Hence, a vehicle not only needs to recognize other road users and other objects, but also comprehend their pose and location to avoid collisions. Participating in real-life road traffic, self-driving vehicles need to gain an absolute understanding of their surroundings. Applications like augmented reality, autonomous driving and other robotic navigation systems are pushing research in this field faster than ever. Gaining a high-level and three-dimensional understanding of digital pictures is one of the major challenges in the field of artificial intelligence.
0 Comments
Leave a Reply. |
Details
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |