FastDVDnet: Towards Real-Time Video Denoising Without Explicit Motion Estimation

Matias Tassano (1, 2) , Julie Delon (3) , Thomas Veit (2)

1 MAP5 - UMR 8145 - Mathématiques Appliquées Paris 5
2 GoPro Technology France
3 MAP5 - UMR 8145 - Mathématiques Appliquées Paris 5

In this paper, we propose a state-of-the-art video denoising algorithm based on a convolutional neural network architecture. Until recently, video denoising with neural networks had been a largely under explored domain, and existing methods could not compete with the performance of the best patch-based methods. The approach we introduce in this paper, called FastDVDnet, shows similar or better performance than other state-of-the-art competitors with significantly lower computing times. In contrast to other existing neural network denoisers, our algorithm exhibits several desirable properties such as fast run-times, and the ability to handle a wide range of noise levels with a single network model. The characteristics of its architecture make it possible to avoid using a costly motion compensation stage while achieving excellent performance. 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.

Mots clés: video denoising; CNN; residual learning; neural networks; image restoration; end-to-end training

Domaines: Traitement des images [eess.IV]; Traitement du signal et de l'image [eess.SP]; Vision par ordinateur et reconnaissance de formes [cs.CV]; Apprentissage [cs.LG]

Matias Tassano, Julie Delon, Thomas Veit. FastDVDnet: Towards Real-Time Video Denoising Without Explicit Motion Estimation. IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Jun 2020, Online, United States. ⟨hal-02170027⟩