Binarized High-Efficiency RAW Video Restoration and Beyond

2026-08-17

AI Summary

This study addresses the limitations of binary neural networks in temporal modeling and activation distribution representation for RAW video restoration by proposing the BinRVR framework. The method introduces a Binary Information Interaction Module (BIIM) to unify spatiotemporal modeling and employs Distribution-Aware Binary Convolution (DAB-Conv), which leverages full-precision statistics to mitigate quantization errors while supporting flexible multi-bit quantization for balancing accuracy and efficiency. Experimental results demonstrate that BinRVR reduces computational cost and parameter count by approximately 96% with only a marginal 4% performance degradation. Furthermore, it achieves superior performance across various RAW video restoration tasks and downstream vision applications, establishing an effective and lightweight solution for high-fidelity video processing.

Abstract

RAW video restoration is fundamental to high-quality low-level perception and serves as the basis for a wide range of downstream vision applications. While binary neural networks (BNNs) enable efficient lightweight deployment for image enhancement, their deficiencies in modeling temporal coherence and activation value distributions hinder their effectiveness when applied to video scenarios. In this paper, we propose BinRVR, a binarized RAW video restoration framework that reduces computation and parameters by approximately 96% while incurring only about 4% performance degradation. Specifically, we present a Binarized Information Interaction Module (BIIM) to jointly model spatial and temporal information in an efficient and unified manner. Moreover, we develop a Distribution-Aware Binarized Convolution (DAB-Conv) that leverages the statistics of full-precision activations to mitigate quantization errors. The proposed framework further supports multi-bit quantization, enabling flexible accuracy-efficiency trade-offs across different hardware constraints. Extensive experiments demonstrate that our BinRVR achieves competitive performance compared with state-of-the-art binarized methods on RAW video restoration tasks, including low-light enhancement, denoising, deblurring, and super-resolution. We further explore the potential of our method on downstream video applications, including object detection and monocular depth estimation.

Problem

• RAW video restoration
• Binary neural networks
• Temporal coherence
• Activation value distribution
• Lightweight deployment

Innovation

• Binarized Neural Networks
• RAW Video Restoration
• Temporal Coherence Modeling
• Distribution-Aware Quantization
• Multi-bit Quantization