Saved in:
Bibliographic Details
Main Authors: Ngo, Tuan Duc, Mirzaei, Ashkan, Qian, Guocheng, Liang, Hanwen, Gan, Chuang, Kalogerakis, Evangelos, Wonka, Peter, Wang, Chaoyang
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2508.01170
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911310610956288
author Ngo, Tuan Duc
Mirzaei, Ashkan
Qian, Guocheng
Liang, Hanwen
Gan, Chuang
Kalogerakis, Evangelos
Wonka, Peter
Wang, Chaoyang
author_facet Ngo, Tuan Duc
Mirzaei, Ashkan
Qian, Guocheng
Liang, Hanwen
Gan, Chuang
Kalogerakis, Evangelos
Wonka, Peter
Wang, Chaoyang
contents We propose a novel algorithm for accelerating dense long-term 3D point tracking in videos. Through analysis of existing state-of-the-art methods, we identify two major computational bottlenecks. First, transformer-based iterative tracking becomes expensive when handling a large number of trajectories. To address this, we introduce a coarse-to-fine strategy that begins tracking with a small subset of points and progressively expands the set of tracked trajectories. The newly added trajectories are initialized using a learnable interpolation module, which is trained end-to-end alongside the tracking network. Second, we propose an optimization that significantly reduces the cost of correlation feature computation, another key bottleneck in prior methods. Together, these improvements lead to a 5-100x speedup over existing approaches while maintaining state-of-the-art tracking accuracy.
format Preprint
id arxiv_https___arxiv_org_abs_2508_01170
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle DELTAv2: Accelerating Dense 3D Tracking
Ngo, Tuan Duc
Mirzaei, Ashkan
Qian, Guocheng
Liang, Hanwen
Gan, Chuang
Kalogerakis, Evangelos
Wonka, Peter
Wang, Chaoyang
Computer Vision and Pattern Recognition
We propose a novel algorithm for accelerating dense long-term 3D point tracking in videos. Through analysis of existing state-of-the-art methods, we identify two major computational bottlenecks. First, transformer-based iterative tracking becomes expensive when handling a large number of trajectories. To address this, we introduce a coarse-to-fine strategy that begins tracking with a small subset of points and progressively expands the set of tracked trajectories. The newly added trajectories are initialized using a learnable interpolation module, which is trained end-to-end alongside the tracking network. Second, we propose an optimization that significantly reduces the cost of correlation feature computation, another key bottleneck in prior methods. Together, these improvements lead to a 5-100x speedup over existing approaches while maintaining state-of-the-art tracking accuracy.
title DELTAv2: Accelerating Dense 3D Tracking
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2508.01170