Scholarlink

<p>We present a modular, production-ready approach that integrates compact Neural Network (NN) into a Kalman-filter-based Multi-Object Tracking (MOT) pipeline. We design three tiny task-specific networks to retain modularity, interpretability and real-time suitability for embedded Automotive Driver Assistance Systems: <OrderedList> <ListItem> <ItemNumber>(i)</ItemNumber> <ItemContent> <p>SPENT (Single-Prediction Network) — predicts per-track states and replaces heuristic motion models used by the Kalman Filter (KF).</p> </ItemContent> </ListItem> <ListItem> <ItemNumber>(ii)</ItemNumber> <ItemContent> <p>SANT (Single-Association Network) — assigns a single incoming sensor object to existing tracks, without relying on heuristic distance and association metrics.</p> </ItemContent> </ListItem> <ListItem> <ItemNumber>(iii)</ItemNumber> <ItemContent> <p>MANTa (Multi-Association Network) — jointly associates multiple sensor objects to multiple tracks in a single step.</p> </ItemContent> </ListItem> </OrderedList> Each module has less than 50k trainable parameters. Furthermore, all three can be operated in real-time, are trained from tracking data, and expose modular interfaces so they can be integrated with standard Kalman-filter state updates and track management. This makes them drop-in compatible with many existing trackers. Modularity is ensured, as each network can be trained and evaluated independently of the others. Our evaluation on the KITTI tracking benchmark shows that SPENT reduces prediction RMSE by more than 50% compared to a standard Kalman filter, while SANT and MANTa achieve up to 95% assignment accuracy. These results demonstrate that small, task-specific neural modules can substantially improve tracking accuracy and robustness without sacrificing modularity, interpretability, or the real-time constraints required for automotive deployment.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Tiny neural networks for multi-object tracking in a modular Kalman framework

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年3月26日

查看全文
  • Christian Holz,
  • Christian Bader,
  • Markus Enzweiler,
  • Matthias Drüppel

摘要

We present a modular, production-ready approach that integrates compact Neural Network (NN) into a Kalman-filter-based Multi-Object Tracking (MOT) pipeline. We design three tiny task-specific networks to retain modularity, interpretability and real-time suitability for embedded Automotive Driver Assistance Systems: (i)

SPENT (Single-Prediction Network) — predicts per-track states and replaces heuristic motion models used by the Kalman Filter (KF).

(ii)

SANT (Single-Association Network) — assigns a single incoming sensor object to existing tracks, without relying on heuristic distance and association metrics.

 (iii)

MANTa (Multi-Association Network) — jointly associates multiple sensor objects to multiple tracks in a single step.

 Each module has less than 50k trainable parameters. Furthermore, all three can be operated in real-time, are trained from tracking data, and expose modular interfaces so they can be integrated with standard Kalman-filter state updates and track management. This makes them drop-in compatible with many existing trackers. Modularity is ensured, as each network can be trained and evaluated independently of the others. Our evaluation on the KITTI tracking benchmark shows that SPENT reduces prediction RMSE by more than 50% compared to a standard Kalman filter, while SANT and MANTa achieve up to 95% assignment accuracy. These results demonstrate that small, task-specific neural modules can substantially improve tracking accuracy and robustness without sacrificing modularity, interpretability, or the real-time constraints required for automotive deployment.