DVDnet: A Fast Network for Deep Video Denoising
About
In this paper, we propose a state-of-the-art video denoising algorithm based on a convolutional neural network architecture. Previous neural network based approaches to video denoising have been unsuccessful as their performance cannot compete with the performance of patch-based methods. However, our approach outperforms other patch-based competitors with significantly lower computing times. In contrast to other existing neural network denoisers, our algorithm exhibits several desirable properties such as a small memory footprint, and the ability to handle a wide range of noise levels with a single network model. The combination between its denoising performance and lower computational load makes this algorithm attractive for practical denoising applications. We compare our method with different state-of-art algorithms, both visually and with respect to objective quality metrics. The experiments show that our algorithm compares favorably to other state-of-art methods. Video examples, code and models are publicly available at \url{https://github.com/m-tassano/dvdnet}.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Video Denoising | Set8 | PSNR36.08 | 136 | |
| Video Denoising | Set8 (test) | PSNR36.08 | 127 | |
| Video Denoising | DAVIS | PSNR38.13 | 79 | |
| Video Denoising | DAVIS 2017 (test) | PSNR38.13 | 60 | |
| Video Denoising | DAVIS 2017 | PSNR38.13 | 51 | |
| Video Denoising | DAVIS (test) | PSNR38.13 | 37 | |
| Video Denoising | IOCV 10 (full) | PSNR38.53 | 13 | |
| Video Denoising | CRVD 87 (full) | PSNR34.5 | 13 | |
| Video Denoising | Set8 960x540 resolution (test) | Inference Time (s)8 | 4 |