SiamAPN++: Siamese Attentional Aggregation Network for Real-Time UAV Tracking

Cao, Ziang; Fu, Changhong; Ye, Junjie; Li, Bowen; Li, Yiming

doi:10.1109/iros51168.2021.9636309

Cited by 84 publications

(42 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Recently, Siamese-based trackers have stood out attributing to their SOTA accuracy and attractive efficiency [2,3,9,26,41,42,78]. For meeting the aerial tracking requirement, some works propose efficient tracking methods [7,21,22].…”

Section: Related Workmentioning

confidence: 99%

“…Recently, the Siamese-based network has emerged as a strong framework for accurate and robust tracking [2,4,11,41,42]. Its efficiency is also optimized in [7,21,22] for the real-time deployment of Siamese-based trackers on aerial platforms. However, the strong correlations inherently existing among consecutive frames, i.e., the temporal information, are neglected by these frameworks, which makes it difficult for these approaches to perceive the motion information of the target objects.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

TCTrack: Temporal Contexts for Aerial Tracking

Cao¹,

Huang²,

Peng³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Temporal contexts among consecutive frames are far from being fully utilized in existing visual trackers. In this work, we present TCTrack 1 , a comprehensive framework to fully exploit temporal contexts for aerial tracking. The temporal contexts are incorporated at two levels: the extraction of features and the refinement of similarity maps. Specifically, for feature extraction, an online temporally adaptive convolution is proposed to enhance the spatial features using temporal information, which is achieved by dynamically calibrating the convolution weights according to the previous frames. For similarity map refinement, we propose an adaptive temporal transformer, which first effectively encodes temporal knowledge in a memory-efficient way, before the temporal knowledge is decoded for accurate adjustment of the similarity map. TCTrack is effective and efficient: evaluation on four aerial tracking benchmarks shows its impressive performance; real-world UAV tests show its high speed of over 27 FPS on NVIDIA Jetson AGX Xavier.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

TCTrack: Temporal Contexts for Aerial Tracking

Cao¹,

Huang²,

Peng³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…SiamAPN [10] designed an anchor proposal network for adaptive anchor proposal. As a continuation of SiamAPN, SiamAPN++ [9] introduces a novel attentional aggregation network to handle semantic information variation. HiFT [11] proposes a brand-new lightweight hierarchical feature transformer for effective and efficient multi-level feature fusion.…”

Section: B Siamese-based Uav Trackingmentioning

confidence: 99%

“…In this section we discuss the effectiveness of our attack methods on SiamRPN++ [24], SiamAPN [10], SiamMask [38], SiamAPN++ [9], SiamGAT [28], SiamCAR [27], and HiFT [11]. Attack results on three public benchmarks are shown in Fig.…”

Section: Overall Performancementioning

confidence: 99%

See 1 more Smart Citation

Ad2Attack: Adaptive Adversarial Attack on Real-Time UAV Tracking

Fu¹,

Li²,

Yuan³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Visual tracking is adopted to extensive unmanned aerial vehicle (UAV)-related applications, which leads to a highly demanding requirement on the robustness of UAV trackers. However, adding imperceptible perturbations can easily fool the tracker and cause tracking failures. This risk is often overlooked and rarely researched at present. Therefore, to help increase awareness of the potential risk and the robustness of UAV tracking, this work proposes a novel adaptive adversarial attack approach, i.e., Ad 2 Attack, against UAV object tracking. Specifically, adversarial examples are generated online during the resampling of the search patch image, which leads trackers to lose the target in the following frames. Ad 2 Attack is composed of a direct downsampling module and a super-resolution upsampling module with adaptive stages. A novel optimization function is proposed for balancing the imperceptibility and efficiency of the attack. Comprehensive experiments on several well-known benchmarks and real-world conditions show the effectiveness of our attack method, which dramatically reduces the performance of the most advanced Siamese trackers.

show abstract