Detection-Free Object Tracking for Multiple Occluded Targets in Plenoptic Video

Yunjeong Yong,Jiwoo Kang,Heeseok Oh

DOI: https://doi.org/10.3390/electronics13030590

IF: 2.9

2024-02-01

Electronics

Abstract:Multiple object tracking (MOT) is a fundamental task in vision, but MOT techniques for plenoptic video are scarce. Almost all 2D MOT algorithms that show high performance mostly use the detection-based method which has the disadvantage of operating only for a specific object. To enable tracking of arbitrary desired objects, this paper introduces a groundbreaking detection-free tracking method for MOT in plenoptic videos. The proposed method deviates from traditional detection-based tracking methods, emphasizing the challenges of tracking targets with occlusions. The paper presents specialized algorithms that exploit the multifocal information of plenoptic video, including the focal range restriction and dynamic focal range adjustment schemes to secure robustness for occluded object tracking. To the improvement of the spatial searching capability, the anchor ensemble and the dynamic change of spatial search region algorithms are also proposed. Additionally, in terms of MOT, to reduce the computation time involved, the motion-adaptive time scheduling technique is proposed, which improves computation speed while guaranteeing a certain level of accuracy. Experimental results show a significant improvement in tracking performance, with a 77% success rate based on intersection over union for occluded targets in plenoptic videos, marking a substantial advancement in the field of plenoptic object tracking.

engineering, electrical & electronic,computer science, information systems,physics, applied

What problem does this paper attempt to address?

The paper attempts to address the problem of detection-free object tracking (DFT) for multiple occluded targets in plenoptic videos. Specifically, existing multi-object tracking (MOT) technologies face the following issues when dealing with plenoptic videos:

1. **Existing technologies mainly focus on 2D videos**: Most existing visual object tracking (VOT) algorithms are limited to 2D videos, with little research utilizing the focal information in plenoptic videos.
2. **Occlusion problem**: In 2D videos, tracking performance significantly degrades when target objects are occluded, as these algorithms rely on detecting specific objects, and occlusion leads to detection failure.
3. **User-defined object tracking**: Existing detection-based tracking methods can only handle specific targets included in the training labels (such as pedestrians, vehicles, etc.), lacking the flexibility to track user-defined objects.

To overcome these issues, the paper proposes a novel detection-free multi-object tracking method that leverages the multi-focal information in plenoptic videos to enhance tracking performance, especially in the presence of occluded targets. The main contributions of the paper include:

1. **Introduction of a detection-free multi-object tracking method**: This method enables tracking of any user-defined objects in plenoptic videos without the need for prior detection.
2. **Proposed focal range limitation and dynamic focal range adjustment algorithms**: These algorithms improve the robustness of tracking occluded targets by limiting and dynamically adjusting the focal range.
3. **Proposed motion-adaptive time scheduling method**: This method dynamically allocates tracking tasks based on the motion speed of objects, enhancing the speed of multi-object tracking while ensuring tracking accuracy.

In summary, the paper aims to develop a more robust and flexible multi-object tracking method by utilizing the multi-focal information in plenoptic videos to address the challenges of occlusion and user-defined object tracking.