# Siamese Attentional Keypoint Network for High Performance Visual Tracking

arXiv: https://arxiv.org/abs/1904.10128 · HTML: https://mlsanity.com/paper/1904.10128/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 23 Apr 2019.

## Built on by (most cited first, 6 of 6)

- [Target Transformed Regression for Accurate Tracking](https://mlsanity.com/paper/2104.00403/) (2021)
- [In Defense and Revival of Bayesian Filtering for Thermal Infrared Object Tracking](https://mlsanity.com/paper/2402.17098/) (2024)
- [Enhancing Visual Representation for Text-based Person Searching](https://mlsanity.com/paper/2412.20646/) (2024)
- [DHRNet: A Dual-Path Hierarchical Relation Network for Multi-Person Pose Estimation](https://mlsanity.com/paper/2404.14025/) (2024)
- [Towards Adaptive Meta-Gradient Adversarial Examples for Visual Tracking](https://mlsanity.com/paper/2505.08999/) (2025)
- [RPT++: Customized Feature Representation for Siamese Visual Tracking](https://mlsanity.com/paper/2110.12194/) (2021)

## Abstract (arXiv)

In this paper, we investigate the impacts of three main aspects of visual tracking, i.e., the backbone network, the attentional mechanism, and the detection component, and propose a Siamese Attentional Keypoint Network, dubbed SATIN, for efficient tracking and accurate localization. Firstly, a new Siamese lightweight hourglass network is specially designed for visual tracking. It takes advantage of the benefits of the repeated bottom-up and top-down inference to capture more global and local contextual information at multiple scales. Secondly, a novel cross-attentional module is utilized to leverage both channel-wise and spatial intermediate attentional information, which can enhance both discriminative and localization capabilities of feature maps. Thirdly, a keypoints detection approach is invented to trace any target object by detecting the top-left corner point, the centroid point, and the bottom-right corner point of its bounding box. Therefore, our SATIN tracker not only has a strong capability to learn more effective object representations, but also is computational and memory storage efficiency, either during the training or testing stages. To the best of our knowledge, we are the first to propose this approach. Without bells and whistles, experimental results demonstrate that our approach achieves state-of-the-art performance on several recent benchmark datasets, at a speed far exceeding 27 frames per second.