EvRWKV: A Continuous Interactive RWKV Framework for Effective Event-Guided Low-Light Image Enhancement

The introduction of the EvRWKV framework marks a significant advancement in low-light image enhancement (LLIE) using event cameras. By addressing the limitations of early and late fusion approaches, EvRWKV employs a Cross-RWKV Module and an Event Image Spectral Fusion Enhancer (EISFE) to maintain feature consistency across different modalities. This continuous interaction allows for improved performance, as evidenced by extensive experiments on the SDE and SDSD datasets, where EvRWKV outperformed traditional image-based methods by 1.79 dB and 1.85 dB in PSNR. Furthermore, the framework's impact on semantic segmentation is notable, with a 35.44% improvement in mean Intersection over Union (mIoU), demonstrating its effectiveness in practical applications. The results highlight the potential of EvRWKV to revolutionize low-light imaging tasks, making it a valuable tool for various fields reliant on high-quality image processing.