Real-Time 3D Single Object Tracking With Transformer

Shan, Jiayao; Zhou, Sifan; Cui, Yubo; Fang, Zheng

doi:10.1109/tmm.2022.3146714

Cited by 30 publications

(8 citation statements)

References 48 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Point Transformer [4], PCT [8], DTNet [35], PVT [41] Object Detection MLCVNet [49], Group-free [50], 3DETR [51], VoTr [42] Object Tracking LTTR [52], PTTR [53], PTT [54] Registration DCP [55], Storm [56], Stickypillars [57], RGM [58], DIT [59] Point Cloud Video Understanding…”

Section: Classification and Segmentationmentioning

confidence: 99%

“…In the field of 3D point cloud processing, high-level tasks usually include: classification & segmentation [4], [7], [8], [28]- [30], [33], [35], [36], [38], [40], [41], [47], [82], [88], [89], object detection [31], [42], [43], [49]- [51], [66], [90], [91] and tracking [52]- [54], registration [55]- [59], [68], [69] and so on. Here, we start by introducing classification & segmentation tasks, which are very common and fundamental research topics in the field of 3D computer vision.…”

Section: D Tasksmentioning

confidence: 99%

“…Unlike the above two approaches which focus on the feature fusion step, Shan et al [54] introduce a Point-Track-Transformer (PTT) module to enhance the feature representation after the feature fusion step. Features from the fusion step and the corresponding point coordinates are mapped into the embedding space.…”

Section: D Tasksmentioning

confidence: 99%

“…As shown in Fig. 2, a number of 3D Transformer backbones are proposed for point cloud classification & segmentation [4], [8], [30], [33], [66], [67], detection [31], [50], tracking [52]- [54], registration [55]- [59], [68], [69], completion [46], [62]- [65], [70] an so on. Moreover, 3D Transformer networks have also been applied to various practical application fields, such as medical data analysis [33] and autonomous driving [71], [72].…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Transformers in 3D Point Clouds: A Survey

Lu¹,

Quan²,

Wei³

et al. 2022

Preprint

View full text Add to dashboard Cite

In recent years, Transformer models have been proven to have the remarkable ability of long-range dependencies modeling. They have achieved satisfactory results both in Natural Language Processing (NLP) and image processing. This significant achievement sparks great interest among researchers in 3D point cloud processing to apply them to various 3D tasks. Due to the inherent permutation invariance and strong global feature learning ability, 3D Transformers are well suited for point cloud processing and analysis. They have achieved competitive or even better performance compared to the state-of-the-art non-Transformer algorithms. This survey aims to provide a comprehensive overview of 3D Transformers designed for various tasks (e.g. point cloud classification, segmentation, object detection, and so on). We start by introducing the fundamental components of the general Transformer and providing a brief description of its application in 2D and 3D fields. Then, we present three different taxonomies (i.e., Transformer implementation-based taxonomy, data representation-based taxonomy, and task-based taxonomy) for method classification, which allows us to analyze involved methods from multiple perspectives. Furthermore, we also conduct an investigation of 3D self-attention mechanism variants designed for performance improvement. To demonstrate the superiority of 3D Transformers, we compare the performance of Transformer-based algorithms in terms of point cloud classification, segmentation, and object detection. Finally, we point out three potential future research directions, expecting to provide some benefit references for the development of 3D Transformers.

show abstract

Section: Classification and Segmentationmentioning

confidence: 99%

Section: D Tasksmentioning

confidence: 99%

Section: D Tasksmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Transformers in 3D Point Clouds: A Survey

Lu¹,

Quan²,

Wei³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Point-Track-Transformer (PTT) [10], [16] creates a transformer module for point-based SOT methods and employs it based on a P2B model, leading to an increased performance.…”

Section: Related Workmentioning

confidence: 99%

VPIT: Real-time Embedded Single Object 3D Tracking Using Voxel Pseudo Images

Oleksiienko¹,

Nousi²,

Passalis³

et al. 2022

Preprint

View full text Add to dashboard Cite

In this paper, we propose a novel voxel-based 3D single object tracking (3D SOT) method called Voxel Pseudo Image Tracking (VPIT). VPIT is the first method that uses voxel pseudo images for 3D SOT. The input point cloud is structured by pillar-based voxelization, and the resulting pseudo image is used as an input to a 2D-like Siamese SOT method. The pseudo image is created in the Bird's-eye View (BEV) coordinates, and therefore the objects in it have constant size. Thus, only the object rotation can change in the new coordinate system and not the object scale. For this reason, we replace multi-scale search with a multi-rotation search, where differently rotated search regions are compared against a single target representation to predict both position and rotation of the object. Experiments on KITTI Tracking dataset show that VPIT is the fastest 3D SOT method and maintains competitive Success and Precision values. Application of a SOT method in a real-world scenario meets with limitations such as lower computational capabilities of embedded devices and a latency-unforgiving environment, where the method is forced to skip certain data frames if the inference speed is not high enough. We implement a real-time evaluation protocol and show that other methods lose most of their performance on embedded devices, while VPIT maintains its ability to track the object.

show abstract

Vpit: real-time embedded single object 3D tracking using voxel pseudo images

Oleksiienko,

Nousi,

Passalis

et al. 2024

Neural Comput & Applic

View full text Add to dashboard Cite

In this paper, we propose a novel voxel-based 3D single object tracking (3D SOT) method called Voxel Pseudo Image Tracking (VPIT). VPIT is the first method that uses voxel pseudo images for 3D SOT. The input point cloud is structured by pillar-based voxelization, and the resulting pseudo image is used as an input to a 2D-like Siamese SOT method. The pseudo image is created in the Bird’s-eye View (BEV) coordinates; and therefore, the objects in it have constant size. Thus, only the object rotation can change in the new coordinate system and not the object scale. For this reason, we replace multi-scale search with a multi-rotation search, where differently rotated search regions are compared against a single target representation to predict both position and rotation of the object. Experiments on KITTI [1] Tracking dataset show that VPIT is the fastest 3D SOT method and maintains competitive Success and Precision values. Application of a SOT method in a real-world scenario meets with limitations such as lower computational capabilities of embedded devices and a latency-unforgiving environment, where the method is forced to skip certain data frames if the inference speed is not high enough. We implement a real-time evaluation protocol and show that other methods lose most of their performance on embedded devices; while, VPIT maintains its ability to track the object.

show abstract

Real-Time 3D Single Object Tracking With Transformer

Cited by 30 publications

References 48 publications

Transformers in 3D Point Clouds: A Survey

Transformers in 3D Point Clouds: A Survey

VPIT: Real-time Embedded Single Object 3D Tracking Using Voxel Pseudo Images

Vpit: real-time embedded single object 3D tracking using voxel pseudo images

Contact Info

Product

Resources

About