Dan Zeng
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Dan Zeng has a total of 14 co-authors. Here are the top co-authors by number of shared publications:
- Shuiwang Li
College of Computer Science and Engineering, Guilin University of Technology, Guilin 541006, China
3 shared publications - Feixiang He
School of Electronic Information, Central South University, Changsha 410083, China
2 shared publications - Hao Zhang
College of Computer Science and Engineering, Guilin University of Technology, Guilin 541006, China
1 shared publication - Hengzhou Ye
Guangxi Key Laboratory of Embedded Technology and Intelligent System, Guilin University of Technology, Guilin, China
1 shared publication - Mengyuan Liu
College of Computer Science and Engineering Guilin University of Technology China
1 shared publication - Pengzhi Zhong1 shared publication
Latest Publications
- Open Access
SAMViTrack: A Search-Region Adaptive Mamba-ViT Tracker for Real-Time UAV Tracking
Achieving fast and robust object tracking is critical for real-time Unmanned Aerial Vehicle (UAV) applications, where targets often move unpredictably and environmental conditions can rapidly change. In this paper, we propose the Search-Region Adaptive Mamba-ViT Tracker (SAMViTrack), a novel framework that combines the efficiency of Mamba attention with the powerful feature extraction capabilities of Vision Transformer (ViT). Our tracker dynamically adjusts the search region based on the target’s motion and environmental context, ensuring precise tracking even under challenging conditions such as occlusions, fast motion, and scale variations. By integrating an adaptive search mechanism, our SAMViTrack significantly reduces computational overhead without compromising accuracy, making it suitable for real-time deployment on UAVs with limited onboard resources. Extensive experiments on benchmark datasets demonstrate that our method outperforms both traditional and modern trackers, achieving superior accuracy and robustness with improved efficiency. The proposed tracker sets a new baseline, especially by combining Mamba and ViT, for UAV tracking by offering a balance between speed, accuracy, and adaptability in dynamic environments.
Open AccessSAMViTrack: A Search-Region Adaptive Mamba-ViT Tracker for Real-Time UAV Tracking
Achieving fast and robust object tracking is critical for real-time Unmanned Aerial Vehicle (UAV) applications, where targets often move unpredictably and environmental conditions can rapidly change. In this paper, we propose the Search-Region Adaptive Mamba-ViT Tracker (SAMViTrack), a novel framework that combines the efficiency of Mamba attention with the powerful feature extraction capabilities of Vision Transformer (ViT). Our tracker dynamically adjusts the search region based on the target’s motion and environmental context, ensuring precise tracking even under challenging conditions such as occlusions, fast motion, and scale variations. By integrating an adaptive search mechanism, our SAMViTrack significantly reduces computational overhead without compromising accuracy, making it suitable for real-time deployment on UAVs with limited onboard resources. Extensive experiments on benchmark datasets demonstrate that our method outperforms both traditional and modern trackers, achieving superior accuracy and robustness with improved efficiency. The proposed tracker sets a new baseline, especially by combining Mamba and ViT, for UAV tracking by offering a balance between speed, accuracy, and adaptability in dynamic environments.
- Open Access
SMTrack: End-to-End Trained Spiking Neural Networks for Multi-Object Tracking in RGB Videos
Brain-inspired Spiking Neural Networks (SNNs) exhibit significant potential for low-power computation, yet their application in visual tasks remains largely confined to image classification, object detection, and event-based tracking. In contrast, real-world vision systems still widely use conventional RGB video streams, where the potential of directly-trained SNNs for complex temporal tasks such as multi-object tracking (MOT) remains underexplored. To address this challenge, we propose SMTrack-the first directly trained deep SNN framework for end-to-end multi-object tracking on standard RGB videos. SMTrack introduces an adaptive and scale-aware Normalized Wasserstein Distance loss (Asa-NWDLoss) to improve detection and localization performance under varying object scales and densities. Specifically, the method computes the average object size within each training batch and dynamically adjusts the normalization factor, thereby enhancing sensitivity to small objects. For the association stage, we incorporate the TrackTrack identity module to maintain robust and consistent object trajectories. Extensive evaluations on BEE24, MOT17, MOT20, and DanceTrack show that SMTrack achieves performance on par with leading ANN-based MOT methods, advancing robust and accurate SNN-based tracking in complex scenarios.
Open AccessSMTrack: End-to-End Trained Spiking Neural Networks for Multi-Object Tracking in RGB Videos
Brain-inspired Spiking Neural Networks (SNNs) exhibit significant potential for low-power computation, yet their application in visual tasks remains largely confined to image classification, object detection, and event-based tracking. In contrast, real-world vision systems still widely use conventional RGB video streams, where the potential of directly-trained SNNs for complex temporal tasks such as multi-object tracking (MOT) remains underexplored. To address this challenge, we propose SMTrack-the first directly trained deep SNN framework for end-to-end multi-object tracking on standard RGB videos. SMTrack introduces an adaptive and scale-aware Normalized Wasserstein Distance loss (Asa-NWDLoss) to improve detection and localization performance under varying object scales and densities. Specifically, the method computes the average object size within each training batch and dynamically adjusts the normalization factor, thereby enhancing sensitivity to small objects. For the association stage, we incorporate the TrackTrack identity module to maintain robust and consistent object trajectories. Extensive evaluations on BEE24, MOT17, MOT20, and DanceTrack show that SMTrack achieves performance on par with leading ANN-based MOT methods, advancing robust and accurate SNN-based tracking in complex scenarios.
- Open Access
Learning Occlusion-Robust Vision Transformers for Real-Time UAV Tracking
Single-stream architectures using Vision Transformer (ViT) backbones show great potential for real-time UAV tracking recently. However, frequent occlusions from obstacles like buildings and trees expose a major drawback: these models often lack strategies to handle occlusions effectively. New methods are needed to enhance the occlusion resilience of single-stream ViT models in aerial tracking. In this work, we propose to learn Occlusion-Robust Representations (ORR) based on ViTs for UAV tracking by enforcing an invariance of the feature representation of a target with respect to random masking operations modeled by a spatial Cox process. Hopefully, this random masking approximately simulates target occlusions, thereby enabling us to learn ViTs that are robust to target occlusion for UAV tracking. This framework is termed ORTrack. Additionally, to facilitate real-time applications, we propose an Adaptive Feature-Based Knowledge Distillation (AFKD) method to create a more compact tracker, which adaptively mimics the behavior of the teacher model ORTrack according to the task's difficulty. This student model, dubbed ORTrack-D, retains much of ORTrack's performance while offering higher efficiency. Extensive experiments on multiple benchmarks validate the effectiveness of our method, demonstrating its state-of-the-art performance. Codes is available at https://github.com/wuyou3474/ORTrack.
Open AccessLearning Occlusion-Robust Vision Transformers for Real-Time UAV Tracking
Single-stream architectures using Vision Transformer (ViT) backbones show great potential for real-time UAV tracking recently. However, frequent occlusions from obstacles like buildings and trees expose a major drawback: these models often lack strategies to handle occlusions effectively. New methods are needed to enhance the occlusion resilience of single-stream ViT models in aerial tracking. In this work, we propose to learn Occlusion-Robust Representations (ORR) based on ViTs for UAV tracking by enforcing an invariance of the feature representation of a target with respect to random masking operations modeled by a spatial Cox process. Hopefully, this random masking approximately simulates target occlusions, thereby enabling us to learn ViTs that are robust to target occlusion for UAV tracking. This framework is termed ORTrack. Additionally, to facilitate real-time applications, we propose an Adaptive Feature-Based Knowledge Distillation (AFKD) method to create a more compact tracker, which adaptively mimics the behavior of the teacher model ORTrack according to the task's difficulty. This student model, dubbed ORTrack-D, retains much of ORTrack's performance while offering higher efficiency. Extensive experiments on multiple benchmarks validate the effectiveness of our method, demonstrating its state-of-the-art performance. Codes is available at https://github.com/wuyou3474/ORTrack.