Organizations: State Key Laboratory of Opto-Electronic Information Acquisition and Protection Technology, the Anhui Provincial Key Laboratory of Secure Artificial Intelligence, Anhui Provincial International Joint Research Center for Advanced Technology in Medical Imaging, and the School of Artificial Intelligence, Anhui University, Hefei, China · Computer Vision Laboratory, University of Sassari, 07100 Sassari, Italy
Event cameras offer a promising sensing modality for face recognition due to their inherent advantages in illumination robustness and privacy-friendliness. However, because event streams lack the stable photometric appearance relied upon by conventional RGB-based face recognition systems, we argue that event-based face recognition should model structure-driven spatiotemporal identity representations shaped by rigid facial motion and individual facial geometry. Since dedicated datasets for event-based face recognition remain lacking, we construct EFace, a small-scale event-based face dataset captured under rigid facial motion. To learn effectively from this limited event data, we further propose EventFace, a framework for event-based face recognition that integrates spatial structure and temporal dynamics for identity modeling. Specifically, we employ Low-Rank Adaptation (LoRA) to transfer structural facial priors from pretrained RGB face models to the event domain, thereby establishing a reliable spatial basis for identity modeling. Building on this foundation, we further introduce a Motion Prompt Encoder (MPE) to explicitly encode temporal features and a Spatiotemporal Modulator (STM) to fuse them with spatial features, thereby enhancing the representation of identity-relevant event patterns. Extensive experiments demonstrate that EventFace achieves the best performance among the evaluated baselines, with a Rank-1 identification rate of 94.19% and an equal error rate (EER) of 5.35%. Results further indicate that EventFace exhibits stronger robustness under degraded illumination than the competing methods. In addition, the learned representations exhibit reduced template reconstructability.
Figures & tables
Fig. 1: (a) Traditional RGB face recognition is sensitive to lighting variations and vulnerable to template reconstruction attacks. (b) By contrast, event-based face recognition leverages structure-driven spatiotemporal patterns generated by rigid facial motion, achieving more stable recognition under varying illumination while reducing reconstructability from leaked templates.
Fig. 2: Overview of the EFace dataset acquisition setup. Subjects perform natural rigid facial motions (e.g., nodding and shaking) to induce intensity changes captured by the event camera, producing spatiotemporal event streams. For the challenging test subset, an RGB camera is mounted in close proximity to the event camera, and the two modalities are acquired in a synchronized manner to ensure consistent illumination conditions.
Fig. 3: Overview of the proposed EventFace framework. The training process is structured into two progressive stages. In Stage I, the pretrained RGB backbone is frozen to retain structural priors, while LoRA modules are trained to align event features with the RGB domain. In Stage II, the adapted backbone is frozen. The Motion Prompt Encoder (MPE) and Spatiotemporal Modulator (STM) are introduced to work collaboratively at each stage: the MPE explicitly extracts motion prompts from event streams, which are then fused with spatial features via the STM module. This mechanism facilitates bidirectional interaction, allowing both spatial and temporal representations to be mutually refined before they are propagated to the next stage.
Fig. 4: Spatial attention visualization via Grad-CAM. Heatmaps highlight regions that contribute most to pairwise similarity.
Fig. 5: Illustration of the Spatiotemporal Interleaved WKV (ST-WKV). Top: The overall data flow. The module takes spatial and motion features as input, concatenates them in an interleaved manner, and processes them through cascaded Bi-WKV layers. Bottom: Visualization of the Row-Major and Column-Major Spatiotemporal Interleaved Scans.
Components
EER (%) ↓
Backbone
Stage I
Stage II
✓
28.81
✓
✓
7.06
✓
✓
10.57
✓
✓
✓
5.35
TABLE I: Effectiveness of the two-stage framework. The backbone is AdaFace pretrained on the WebFace4M dataset.
Strategy
Rank-1 (%) ↑
EER (%) ↓
Full Fine-Tuning
91.18
7.26
Adapter Tuning
90.73
7.96
LoRA Tuning
92.09
6.81
TABLE II: Performance comparison of different tuning strategies.
MPE
STM
Rank-1 (%) ↑
EER (%) ↓
Kernel Size
Token Shift
Token Arrangement
5-3
Octa-Shift
Interleaved
93.65
5.82
7-3
Q-Shift
Interleaved
93.20
6.45
7-3
Octa-Shift
Sequential
92.24
6.18
7-3
Octa-Shift
Interleaved
94.19
5.35
TABLE III: Joint ablation of motion encoding and spatiotemporal modeling strategies in Stage II.
Fig. 6: Effect of the input event-frame count on recognition performance and computational cost. Each event frame is accumulated over a fixed 50 ms interval.
Fig. 7: CMC, DET, and ROC curves of EventFace and competing methods on the EFace benchmark. EventFace achieves the best overall identification and verification performance among the evaluated methods.
Fig. 8: Cosine similarity distributions under the degraded illumination challenge. (a) Event modality and (b) RGB modality. The plots illustrate the distributions of genuine (green) and impostor (gray) matching pairs.
Fig. 9: Visualization of matching examples under the degraded illumination challenge. The Gallery samples are captured under standard indoor lighting, while the Probe samples correspond to varying low-light levels (Levels 1–3) and sidelight conditions. The computed cosine similarity scores are annotated for both RGB and Event modalities.
Fig. 10: Visualization of face reconstruction from leaked feature templates. The top rows (GT) display the ground-truth Event Frames, while the bottom rows (Shahreza et al. ) show the reconstructed results.
Fig. 11: Privacy evaluation under a worst-case raw data leakage scenario. We employ representative event-to-video reconstruction algorithms, E2VID and Sparse-E2VID, to recover visual information directly from the 200 ms raw event streams.