An end-to-end multi-target tracking method and system based on hierarchical spatio-temporal memory and cross-modal prior

By employing an end-to-end multi-target tracking method based on hierarchical spatiotemporal memory and cross-modal priors, the problem of unstable target trajectories in complex scenarios is solved, achieving efficient and adaptive target tracking and improving trajectory continuity and identity consistency.

CN122156255APending Publication Date: 2026-06-05BEIJING INST OF TECH
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN Β· China
Patent Type
Applications(China)
Current Assignee / Owner
BEIJING INST OF TECH
Filing Date
2026-03-20
Publication Date
2026-06-05

AI Technical Summary

Technical Problem

Existing technologies struggle to achieve stable correlation of target trajectories and identity preservation in complex scenarios, especially under the influence of factors such as camera shake, changes in target pose, occlusion, and changes in lighting, leading to trajectory drift and identity loss. Furthermore, tracking performance degrades when visual information is insufficient.

Method used

An end-to-end multi-target tracking method based on hierarchical spatiotemporal memory and cross-modal priors is adopted. By constructing short-term memory modules and long-term memory modules, and combining cross-attention mechanisms and localization prior information from multimodal sensors, the trajectory state output is optimized.

Benefits of technology

It significantly improves the robustness and accuracy of target tracking in complex scenarios, suppresses instantaneous identity switching and trajectory breakage, and ensures stable tracking continuity when visual information is insufficient.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122156255A_ABST
    Figure CN122156255A_ABST
Patent Text Reader

Abstract

The application discloses an end-to-end multi-target tracking method and system based on layered space-time memory and cross-modal priori. The application cooperatively deals with the problems of instantaneous instability and long-time appearance mutation in the tracking process by constructing a short-term memory module and a long-term memory module. The short-term memory adopts a ring buffer structure to cache recent trajectory representation, and provides continuous association clues through time proximity and geometric accessibility screening; the long-term memory adopts a novelty-driven strategy to maintain a sparse appearance prototype library, and provides diversified appearance anchor points for occlusion reappearance and re-identification. Further, the method integrates multi-modal positioning priori, encodes it into a feature sequence through space-time alignment and coordinate unification, and performs cross-modal attention fusion with the visual dominant trajectory query. The method unifies the short-term continuity, long-term recognition and absolute position constraint in an end-to-end framework, effectively improving the identity consistency and tracking continuity of target trajectories in complex scenes.
Need to check novelty before this filing date? Find Prior Art