Saved in:
Bibliographic Details
Main Authors: Duché, Axel, Chatelain, Clément, Gasso, Gilles
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2602.00153
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917236394950656
author Duché, Axel
Chatelain, Clément
Gasso, Gilles
author_facet Duché, Axel
Chatelain, Clément
Gasso, Gilles
contents We propose a lightweight compressed-domain tracking model that operates directly on video streams, without requiring full RGB video decoding. Using motion vectors and transform coefficients from compressed data, our deep model propagates object bounding boxes across frames, achieving a computational speed-up of order up to 3.7 with only a slight 4% mAP@0.5 drop vs RGB baseline on MOTS15/17/20 datasets. These results highlight codec-domain motion modeling efficiency for real-time analytics in large monitoring systems.
format Preprint
id arxiv_https___arxiv_org_abs_2602_00153
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle See Without Decoding: Motion-Vector-Based Tracking in Compressed Video
Duché, Axel
Chatelain, Clément
Gasso, Gilles
Computer Vision and Pattern Recognition
Artificial Intelligence
Image and Video Processing
We propose a lightweight compressed-domain tracking model that operates directly on video streams, without requiring full RGB video decoding. Using motion vectors and transform coefficients from compressed data, our deep model propagates object bounding boxes across frames, achieving a computational speed-up of order up to 3.7 with only a slight 4% mAP@0.5 drop vs RGB baseline on MOTS15/17/20 datasets. These results highlight codec-domain motion modeling efficiency for real-time analytics in large monitoring systems.
title See Without Decoding: Motion-Vector-Based Tracking in Compressed Video
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Image and Video Processing
url https://arxiv.org/abs/2602.00153