mlsanityData to 14 Oct 2026

MVTrack: Ultrafast Appearance-Free Moving Object Tracking from Compressed Bitstreams

arXiv 2608.10790 ↗First posted 11 Aug 2026cs.CV cs.AIcited by 0

Not on the map. Why?

Abstract (arXiv)

Deploying modern video trackers at scale is bottlenecked by the computational cost of RGB-based object detectors. To this end, we present MVTrack, an ultrafast tracker for moving objects that operates directly on H.264 bitstreams. MVTrack combines MVDet, a lightweight detector for motion vector fields, with MVLink, a minimalist kinematic association module. On VIRAT, MVTrack outperforms YOLO26n while using 60× fewer parameters, requiring 40× fewer FLOPs, and reducing CPU latency by 8.6×. These results demonstrate that compressed video data alone can enable accurate and scalable surveillance tracking, thereby bypassing the need for pixel reconstruction.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 26 of its 49 references that are in the map. First posted on arXiv 11 Aug 2026.

Builds on

26 of 49 references.

References · 26

  1. 2014Adam: A Method for Stochastic Optimizationcited by 37,240
  2. 2014Microsoft COCO: Common Objects in Contextcited by 17,608
  3. 2019Optuna: A Next-generation Hyperparameter Optimization Frameworkcited by 2,501
  4. 2019Objects as Pointscited by 1,175
  5. 2021ByteTrack: Multi-Object Tracking by Associating Every Detection Boxcited by 823
  6. 2016Performance Measures and a Data Set for Multi-Target, Multi-Camera Trackingcited by 692
  7. 2020HOTA: A Higher Order Metric for Evaluating Multi-Object Trackingcited by 552
  8. 2015Delving Deeper into Convolutional Networks for Learning Video Representationscited by 148
  9. 2017Compressed Video Action Recognitioncited by 106
  10. 2020Enhancing Geometric Factors in Model Learning and Inference for Object Detection and Instance Segmentationcited by 98
  11. 2016Real-time Action Recognition with Enhanced Motion Vector CNNscited by 51
  12. 2018Fast Object Detection in Compressed Videocited by 33
14 more
  1. 2025Efficient Motion-Aware Video MLLMcited by 14
  2. 2026CoPE-VideoLM: Leveraging Codec Primitives For Efficient Video Language Modelingcited by 11
  3. 2021Accelerating Video Object Segmentation with Compressed Videocited by 10
  4. 2023Efficient Semantic Segmentation by Altering Resolutions for Compressed Videoscited by 8
  5. 2023MVFlow: Deep Optical Flow Estimation of Compressed Videos with Motion Vector Priorcited by 7
  6. 2018MV-YOLO: Motion Vector-aided Tracking by Semantic Object Detectioncited by 6
  7. 2022Real-time Online Multi-Object Tracking in Compressed Domaincited by 6
  8. 2022CoVA: Exploiting Compressed-Domain Analysis to Accelerate Video Analyticscited by 6
  9. 2021Motion Vector Extrapolation for Video Object Detectioncited by 5
  10. 2026See Without Decoding: Motion-Vector-Based Tracking in Compressed Videocited by 4
  11. 2017Can you tell a face from a HEVC bitstream?cited by 3
  12. 2023Efficient Temporally-Aware DeepFake Detection using H.264 Motion Vectorscited by 3
  13. 2025MoCLIP-Lite: Efficient Video Recognition by Fusing CLIP with Motion Vectorscited by 2
  14. 2025MVP: Motion Vector Propagation for Zero-Shot Video Object Detectioncited by 2

Built on by

Not cited yet.