轻量级视频超分模型，可在边缘设备实时运行。
NanoVSR: Towards Real-Time Video Super-Resolution on Edge Devices
- 全卷积结构+结构重参数化，推理时转为标准卷积
- 28.64 dB PSNR，16GB Jetson Orin上达27.2 FPS
- 无需显式运动补偿，适合部署在边缘硬件
现有视频超分辨率（VSR）方法依赖变换器和显式光流，导致计算开销大且不兼容硬件加速器（如TensorRT）。为此，我们提出NanoVSR，一种专为资源受限边缘设备设计的可扩展全卷积架构。通过结构重参数化，NanoVSR在推理阶段坍缩为标准卷积，确保与硬件无缝兼容且运行时开销可忽略。尽管未采用显式运动补偿，它仍通过渐进式训练隐式学习时空对齐，保持了优异的重建质量。在REDS4基准上，NanoVSR展现出精度与效率的出色平衡，显著优化紧凑架构的权衡。其基线模型NanoVSR-644k在NVIDIA Jetson Orin NX 16GB（25W）上实现28.64 dB PSNR和27.2 FPS；扩展版NanoVSR-1.7M达到29.15 dB，吞吐量19.58 FPS，提供卓越的边缘优化超分能力。代码已开源：https://github.com/filippawlicki/nanovsr。
原文摘要 · Abstract (English)
Recent Video Super-Resolution (VSR) methods rely heavily on transformers and explicit optical flow, creating computational overhead and custom operations that hinder deployment on hardware accelerators like TensorRT. To address this, we introduce NanoVSR, a scalable, fully convolutional architecture designed for resource-constrained edge devices. Using structural reparameterization, NanoVSR collapses into standard convolutions during inference, ensuring seamless hardware compatibility and negligible runtime overhead. Furthermore, despite lacking explicit motion compensation, it maintains competitive restoration quality by implicitly learning spatio-temporal alignments through progressive training. Evaluated on the REDS4 benchmark, NanoVSR demonstrates an exceptional balance between accuracy and computational efficiency, significantly improving the trade-off for compact architectures. Our NanoVSR-644k baseline yields 28.64 dB PSNR while delivering 27.2 FPS on the NVIDIA Jetson Orin NX 16GB (25W), offering massive speed gains over heavier models. The scaled NanoVSR-1.7M variant reaches 29.15 dB with a throughput of 19.58 FPS, providing superior, edge-optimized upscaling. Code is available at https://github.com/filippawlicki/nanovsr.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。