SpikeTrack: A Spike-driven Framework for Efficient Visual Tracking

Qiuyang Zhang; Jiujun Cheng; Qichao Mao; Cong Liu; Yu Fang; Yuhong Li; Mengying Ge; Shangce Gao

cs.CVFeb 27, 2026

SpikeTrack: A Spike-driven Framework for Efficient Visual Tracking

Authors: Qiuyang Zhang, Jiujun Cheng, Qichao Mao, Cong Liu, Yu Fang, Yuhong Li, Mengying Ge, Shangce Gao

Organizations: Tongji University · Nova University of Lisbon · Stockholm University · Shanghai University ·

Abstract

Spiking Neural Networks (SNNs) promise energy-efficient vision, but applying them to RGB visual tracking remains difficult: Existing SNN tracking frameworks either do not fully align with spike-driven computation or do not fully leverage neurons' spatiotemporal dynamics, leading to a trade-off between efficiency and accuracy. To address this, we introduce SpikeTrack, a spike-driven framework for energy-efficient RGB object tracking. SpikeTrack employs a novel asymmetric design that uses asymmetric timestep expansion and unidirectional information flow, harnessing spatiotemporal dynamics while cutting computation. To ensure effective unidirectional information transfer between branches, we design a memory-retrieval module inspired by neural inference mechanisms. This module recurrently queries a compact memory initialized by the template to retrieve target cues and sharpen target perception over time. Extensive experiments demonstrate that SpikeTrack achieves the state-of-the-art among SNN-based trackers and remains competitive with advanced ANN trackers. Notably, it surpasses TransT on LaSOT dataset while consuming only 1/26 of its energy. To our knowledge, SpikeTrack is the first spike-driven framework to make RGB tracking both accurate and energy efficient. The code and models are available at https://github.com/faicaiwawa/SpikeTrack.

Figures & tables

Figure 1: Energy–accuracy trade-off on LaSOT [ 6 ] . SpikeTrack achieves lower energy consumption than efficient ANN trackers while matching the accuracy of precision-oriented methods.

Figure 2: Structure comparison between one-stream tracking SNN (top) and our asymmetric tracking SNN (bottom). L represents the number of blocks in the backbone.

Figure 3: Overview of SpikeTrack. The network consists of three components: a weight-sharing siamese backbone, a memory retrieval module for information transfer, and a prediction head. We use asymmetric timestep inputs and unidirectional information flow. During inference, template branch features are converted and cached as memory. The search branch queries this memory to extract target cues. The template branchs runs once, per initialization or update.

Figure 4: Implementation details of the Memory Retrieval Module. The purple legend (bottom left) illustrates the recurrent, looped connectivity structure in the brain. For simplicity of illustration, the temporal spiking across timesteps are omitted.

	Method	Efficiency			TrackingNet			GOT-10k			LaSOT			LaSOT ext
	Method	Para.	Pow.	T $\times$ D	AUC	P N	P	AO	SR 50	SR 75	AUC	P N	P	AUC	P N	P
SNN	SpikeTrack-B 384	36.8	27.3	3 $\times$ 4	82.0	87.6	80.7	73.1	84.0	69.9	66.7	76.8	72.9	47.6	58.5	54.5
	SpikeTrack-B 384		20.0	1 $\times$ 4	81.8	87.1	80.4	70.7	81.0	67.0	67.5	77.8	73.3	47.3	57.8	54.1
	SpikeTrack-B 256	36.8	9.8	3 $\times$ 4	81.1	86.9	79.1	72.2	83.9	67.7	67.1	77.7	72.5	46.7	57.6	52.9
	SpikeTrack-B 256		8.1	1 $\times$ 4	80.1	85.4	77.6	69.6	80.6	66.2	66.6	77.2	71.6	46.0	56.7	52.0
	SpikeTrack-S 256	11.2	3.7	3 $\times$ 4	78.7	84.6	75.5	67.8	79.9	61.7	64.5	76.0	69.0	43.9	54.5	49.2

Table 1: Comparison of performance and efficiency across tracking methods. Para. and Pow. denote parameters (M) and power (

mJ

), respectively; T and D represent the timestep and maximum integer value emitted during training. Results marked with ∗ are trained solely on the GOT-10K set. All results are reported in percentage (%). The top two snn results are highlighted in bold and underlined , respectively.

Method	T $\times$ D	Pow.	TNL2K	OTB	UAV
SpikeTrack-B 384	3 $\times$ 4	27.3	54.8	68.9	68.3
SpikeTrack-B 384	1 $\times$ 4	20.0	54.2	68.3	67.2
SpikeTrack-B 256	3 $\times$ 4	9.8	54.0	69.6	67.8
SpikeTrack-B 256	1 $\times$ 4	8.1	53.3	68.0	67.5
SpikeTrack-S 256	3 $\times$ 4	3.7	52.0	69.4	66.2
SpikeTrack-S 256	1 $\times$ 4	2.8	51.2	67.1	66.3

Table 2: Performance comparison of different tracking methods. All evaluation metrics are based on Success Rate (AUC) and reported in percentage (%). The top two snn results are highlighted in bold and underlined , respectively.

#	Method	Pow.	GOT-10K	LaSOT	$\bigtriangleup$
1	Baseline	8.7	71.3	66.8	-
2	Fine-tuning	8.7	71.6	66.9	+0.2
3	One-stream	22.8	70.8	65.4	-0.8
4a	Vanilla Cross-attn	7.6	70.9	65.0	-1.1
4b	Modulation	6.8	58.3	49.9	-14.9
4c	Modulation ∗	10.2	64.0	58.1	-8.0

Table 3: Ablation Study on LaSOT and GOT-10k.

\bigtriangleup

denotes the performance change (averaged over benchmarks) compared with the baseline. ∗ indicates the non-spike version.

Figure 5: Influence of the number of retrieval loop in MRM.

Figure 6: Gap analysis between SpikeTrack and a precision-oriented ANN across LaSOT attributes.

Figure 7: Visualization of the spike tensor produced by MRM. Three cases are shown: similar objects, occlusion, and background interference.

Figure 8: Visualization of the spike tensor obtained after two retrievals in MRM, based on spike firing rate.

Figure 9: Visualization comparison of SpikeTrack and other ANN-based Trackers.

Figure 10: Performance of SpikeTrack and other ANN-based Trackers on the LaSOT attribute challenges.

GOT	25	35	45	55
0.65	72.2	72.5	72.2	72.1
0.7	72.2	72.2	72.0	71.9
0.75	71.8	72.4	71.6	71.8
0.8	71.7	72.3	71.3	71.7

Table 4: Template update parameter sensitivity analysis of SpikeTrack-B 256 -T3. Rows indicate update confidence and columns indicate update interval. Underlines indicate default unified settings of the SpikeTrack Family for short- and long-term datasets.

		ANN-based Tracker	Spike-driven Tracker
Conv	First Conv	$E_{MAC}\cdot FL_{Conv}$	$E_{MAC}\cdot T\cdot R\cdot FL_{Conv}$
Conv	Other Conv	$E_{MAC}\cdot FL_{Conv}$	$E_{AC}\cdot T\cdot R\cdot FL_{Conv}$
Self-Attention	$Q,K,V$	$E_{MAC}\cdot 3ND^{2}$	$E_{AC}\cdot T\cdot R\cdot 3FL_{Conv}$
	$f(Q,K,V)$	$E_{MAC}\cdot 2N^{2}D$	$E_{AC}\cdot T\cdot R\cdot 2ND^{2}$
	Scale	$E_{M}\cdot N^{2}$	-
	Softmax	$E_{MAC}\cdot 2N^{2}$	-

Table 5: Energy evaluation.

FL_{Conv}

and

FL_{MLP}

represent the FLOPs of the Conv and MLP models in the ANNs, repectively.

R

denote the spiking firing rate (the proportion of non-zero elements in the spike matrix) of the layer corresponding to the operator.

T

is timestep.

E_{MAC}

and

E_{AC}

are the energy consumption of performing one MAC and one AC operation, respectively. In the spike-driven attention, the scale operation is avoided by incorporating the scale factor into the neuron’s leakage factor. The first downsampling layer uses

MAC

convolution, consistent with other works [ 35 , 36 , 41 , 26 , 34 ] .

				T1	T2	T3
Stage 1	DownSampling		Conv	1	-	-
Stage 1	ConvFormer Spike Block	SepSpikeConv	PWConv1	1.4643	-	-
			DWConv	1.3569	-	-
			PWConv2	1.2556	-	-
		Channel Conv	Conv1	1.5301	-	-
		Channel Conv	Conv2	0.2497	-	-

Table 6: Layer spiking firing rates of SpikeTrack-B256-T3 Search Branch.

				T1	T2	T3
Stage 1	DownSampling		Conv	1	1	1
Stage 1	ConvFormer Spike Block	SepSpikeConv	PWConv1	1.458	1.4853	1.5083
			DWConv	1.3534	1.3790	1.3954
			PWConv2	1.2563	1.3188	1.3337
		Channel Conv	Conv1	1.5297	1.5035	1.5087
		Channel Conv	Conv2	0.2500	0.2496	0.2460

Table 7: Layer spiking firing rates of SpikeTrack-B256-T3 Template Branch.

Visual Object Tracking

Visual object tracking follows objects across video frames, maintaining their locations and identities despite motion, occlusion, and changing appearance.

See all 175 papers

STATrack: A Target-Aware Fully Spiking Neural Network for Efficient UAV Tracking
Mar 29, 2026Pengzhi Zhong, Jiwei Mo, Dan Zeng +2Visual Object Tracking Spiking Neural Networks
Guilin University of Technology, Guilin 541006, China · School of Artificial Intelligence, Sun Yat-sen University, Zhuhai 510275, China · School of Electronic Information, Central South University, Changsha, Hunan, 410083, China
SpikingMOT: A Spike-Driven Multi-Object Tracker
Jul 22, 2026Yiding Sun, Xiangyang Yang, Dongxu Zhang +7Activation Sparsity Spiking Neural Networks
1Xi’an Jiaotong University · 2Peking University · 3Guilin University of Technology
SBMVTrack: Spike-Budgeted Multi-View Learning for Power-Efficient UAV Tracking
Sep 21, 2026Pengzhi Zhong, Jiwei Mo, Haolun Li +4Activation Sparsity Visual Object Tracking
College of Computer Science and Engineering, Guilin University of Technology, Guilin 541004, China · School of Information Engineering, Wuhan University of Technology, Wuhan 430070, China

Mar 29, 2026J/K move · Enter open · S save

STATrack: A Target-Aware Fully Spiking Neural Network for Efficient UAV Tracking

Pengzhi Zhong, Jiwei Mo, Dan Zeng, Feixiang He+1

Guilin University of Technology, Guilin 541006, China · School of Artificial Intelligence, Sun Yat-sen University, Zhuhai 510275, China · School of Electronic Information, Central South University, Changsha, Hunan, 410083, China

Visual Object Tracking Spiking Neural Networks

Open paper AskPDF arXiv

Spiking Neural Networks (SNNs), characterized by their event-driven computation and low power consumption, have shown great potential for energy-efficient visual tracking on unmanned aerial vehicles (UAVs). However, existing SNN-based trackers often rely on costly event cameras, which limits their deployment on standard RGB-camera UAV platforms. To address this limitation, we propose STATrack, a fully spiking neural network framework for UAV visual tracking using only RGB inputs. To the best of our knowledge, this is the first study to explore fully spiking neural networks for RGB-based UAV visual tracking. Considering the challenge of preserving fine-grained target information in deep spiking representations under a limited number of time steps, we introduce Adaptive Mutual Information Maximization (AMIM). AMIM maximizes the statistical dependency between target-centered templates and their deep spiking representations, encouraging the spiking backbone to learn representations that better distinguish the target from surrounding backgrounds. We further develop a sample-difficulty-aware dynamic weighting strategy that adaptively adjusts the mutual-information constraint according to the relative localization difficulty of each training batch. Extensive experiments on four widely used UAV tracking benchmarks demonstrate that STATrack achieves state-of-the-art tracking performance with low theoretical energy consumption, highlighting its potential for energy-constrained UAV applications. Code is released at: https://anonymous.4open.science/r/STATrack.

Abstract

Visual Object Tracking

Explore similar work

STATrack: A Target-Aware Fully Spiking Neural Network for Efficient UAV Tracking

SpikingMOT: A Spike-Driven Multi-Object Tracker

SBMVTrack: Spike-Budgeted Multi-View Learning for Power-Efficient UAV Tracking