Energy-efficient spiking neural network improves drone visual tracking
SBMVTrack: Spike-Budgeted Multi-View Learning for Energy-Efficient UAV Tracking
Computer Vision and Pattern Recognition
Summary
Tracking objects with drones requires fast and energy-saving technology to keep the battery alive longer. The authors present SBMVTrack, a new system that controls the brain-like spikes used by the drone's tracking AI to save energy. They make the system smarter by letting different views of the target help each other, so the drone better recognizes what to follow while using less power. Tests show SBMVTrack keeps tracking accuracy high while cutting down energy use.
What this means in practice
- •For drone engineers: Design UAVs that track objects more energy-efficiently by optimizing neural spike activity within tracking systems under computational constraints.
- •For embedded systems developers: Integrate energy-weighted spike budgeting methods into low-power devices requiring visual tracking to reduce computational energy use while maintaining accuracy.
Authors
Pengzhi Zhong, Jiwei Mo, Haolun Li, Ge Zheng, Jingqi Wang, Xinyi Bo, Shuiwang Li
Abstract
With sparse and event-driven computation, spiking neural networks show great potential for achieving accurate and energy-efficient UAV visual tracking. However, existing SNN-based trackers typically use spike firing rates only for energy evaluation and lack explicit optimization of actual spike activity. To address this, we propose SBMVTrack, a fully spiking framework for energy-efficient UAV tracking. SBMVTrack introduces Energy-Weighted Spike Budgeting (EWSB). EWSB weights actual spike activity according to the computational cost of each spiking layer. It constrains the energy-weighted firing rate and saturation activity, thereby reducing redundant spike computations. To improve tracking performance under the spike budget constraint, we propose Masked Multi-View Target Modeling (MVTM). This method treats the initial template, online template, and search region from the same sequence as correlated temporal views. It enhances the robustness of target representations through cross-view feature completion and identity-consistency learning. Extensive experiments on multiple benchmarks demonstrate that SBMVTrack effectively reduces the average spike firing rate and theoretical energy consumption. Meanwhile, it maintains competitive tracking performance, achieving a better accuracy-energy trade-off. The source code will be released upon acceptance.