Dynamic roi target closed loop tracking method, system and related devices
By employing a heterogeneous hardware architecture combining the STM32N6 main control module and the FPGA coprocessor module, along with a Kalman filter algorithm, real-time stable target tracking of ultra-high-definition video on low-power miniaturized edge devices was achieved. This resolved the contradiction between computing power and real-time performance, as well as the target stability issue in existing technologies, thereby improving tracking accuracy and stability.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- BAO NA SHENG (SHENZHEN) TECH CO LTD
- Filing Date
- 2026-04-21
- Publication Date
- 2026-07-17
AI Technical Summary
Existing technologies struggle to achieve real-time, stable target tracking of ultra-high-definition video on low-power, miniaturized edge devices. This is due to issues such as the conflict between computing power and real-time performance, hardware architecture bottlenecks, insufficient target centering stability, and poor compatibility with power consumption and miniaturization.
A heterogeneous hardware architecture combining an STM32N6 main control module and an FPGA coprocessor module is adopted. The FPGA coprocessor module performs image cropping and scaling, while the STM32N6 main control module performs AI target inference and trajectory prediction, constructing a dynamic ROI closed-loop tracking link. Combined with the Kalman filter algorithm, delay compensation and ROI adaptive adjustment are performed to achieve efficient target tracking.
It achieves real-time stable target tracking of ultra-high-definition video under low power consumption, solves the computing power and bandwidth bottleneck of edge devices, ensures dynamic tracking of the target in the ROI center, and improves tracking stability in high-speed movement and near-far zoom scenarios.
Smart Images
Figure CN122415982A_ABST