# Multi-modality Deep Restoration of Extremely Compressed Face Videos

arXiv: https://arxiv.org/abs/2107.05548 · HTML: https://mlsanity.com/paper/2107.05548/

## Key facts

- Cited by 14 papers in the map (data through 14 Oct 2026).
- Builds on 42 of its 74 references that are in the map.
- First posted on arXiv 5 Jul 2021.

## Builds on (most cited first, 30 of 42)

- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [Deformable Convolutional Networks](https://mlsanity.com/paper/1703.06211/) (2017)
- [VoxCeleb2: Deep Speaker Recognition](https://mlsanity.com/paper/1806.05622/) (2018)
- [VoxCeleb: a large-scale speaker identification dataset](https://mlsanity.com/paper/1706.08612/) (2017)
- [Video Enhancement with Task-Oriented Flow](https://mlsanity.com/paper/1711.09078/) (2017)
- [EDVR: Video Restoration with Enhanced Deformable Convolutional Networks](https://mlsanity.com/paper/1905.02716/) (2019)
- [DVC: An End-to-end Deep Video Compression Framework](https://mlsanity.com/paper/1812.00101/) (2018)
- [MakeItTalk: Speaker-Aware Talking-Head Animation](https://mlsanity.com/paper/2004.12992/) (2020)
- [Pose-Controllable Talking Face Generation by Implicitly Modularized Audio-Visual Representation](https://mlsanity.com/paper/2104.11116/) (2021)
- [Few-Shot Adversarial Learning of Realistic Neural Talking Head Models](https://mlsanity.com/paper/1905.08233/) (2019)
- [Hierarchical Cross-Modal Talking Face Generationwith Dynamic Pixel-Wise Loss](https://mlsanity.com/paper/1905.03820/) (2019)
- [Neural Voice Puppetry: Audio-driven Facial Reenactment](https://mlsanity.com/paper/1912.05566/) (2019)
- [TDAN: Temporally Deformable Alignment Network for Video Super-Resolution](https://mlsanity.com/paper/1812.02898/) (2018)
- [Talking Face Generation by Adversarially Disentangled Audio-Visual Representation](https://mlsanity.com/paper/1807.07860/) (2018)
- [X2Face: A network for controlling face generation by using images, audio, and pose codes](https://mlsanity.com/paper/1807.10550/) (2018)
- [FSRNet: End-to-End Learning Face Super-Resolution with Facial Priors](https://mlsanity.com/paper/1711.10703/) (2017)
- [Compression Artifacts Reduction by a Deep Convolutional Network](https://mlsanity.com/paper/1504.06993/) (2015)
- [M-LVC: Multiple Frames Prediction for Learned Video Compression](https://mlsanity.com/paper/2004.10290/) (2020)
- [You said that?](https://mlsanity.com/paper/1705.02966/) (2017)
- [Video Compression With Rate-Distortion Autoencoders](https://mlsanity.com/paper/1908.05717/) (2019)
- [Learned Video Compression](https://mlsanity.com/paper/1811.06981/) (2018)
- [Learning Warped Guidance for Blind Face Restoration](https://mlsanity.com/paper/1804.04829/) (2018)
- [Deep Semantic Face Deblurring](https://mlsanity.com/paper/1803.03345/) (2018)
- [Progressive Face Super-Resolution via Attention to Facial Landmark](https://mlsanity.com/paper/1908.08239/) (2019)
- [VisemeNet: Audio-Driven Animator-Centric Speech Animation](https://mlsanity.com/paper/1805.09488/) (2018)
- [Improving Deep Video Compression by Resolution-adaptive Flow Coding](https://mlsanity.com/paper/2009.05982/) (2020)
- [Exemplar Guided Face Image Super-Resolution without Facial Landmarks](https://mlsanity.com/paper/1906.07078/) (2019)
- [Multi-Frame Quality Enhancement for Compressed Video](https://mlsanity.com/paper/1803.04680/) (2018)
- [Deep Generative Adversarial Compression Artifact Removal](https://mlsanity.com/paper/1704.02518/) (2017)
- [Deep Cascaded Bi-Network for Face Hallucination](https://mlsanity.com/paper/1607.05046/) (2016)

## Built on by (most cited first, 14 of 14)

- [Dual Associated Encoder for Face Restoration](https://mlsanity.com/paper/2308.07314/) (2023)
- [LVQAC: Lattice Vector Quantization Coupled with Spatially Adaptive Companding for Efficient Learned Image Compression](https://mlsanity.com/paper/2304.12319/) (2023)
- [Learning Optimal Lattice Vector Quantizers for End-to-end Neural Image Compression](https://mlsanity.com/paper/2411.16119/) (2024)
- [Perceptual Quality Improvement in Videoconferencing using Keyframes-based GAN](https://mlsanity.com/paper/2311.04263/) (2023)
- [Compressed Deepfake Video Detection Based on 3D Spatiotemporal Trajectories](https://mlsanity.com/paper/2404.18149/) (2024)
- [Dual-layer Image Compression via Adaptive Downsampling and Spatially Varying Upconversion](https://mlsanity.com/paper/2302.06096/) (2023)
- [Receptive Field Expanded Look-Up Tables for Vision Inference: Advancing from Low-level to High-level Tasks](https://mlsanity.com/paper/2510.10522/) (2025)
- [FLLIC: Functionally Lossless Image Compression](https://mlsanity.com/paper/2401.13616/) (2024)
- [Audio-Assisted Face Video Restoration with Temporal and Identity Complementary Learning](https://mlsanity.com/paper/2508.04161/) (2025)
- [HeadsUp! High-Fidelity Portrait Image Super-Resolution](https://mlsanity.com/paper/2510.09924/) (2025)
- [Anisotropic Pooling for LUT-realizable CNN Image Restoration](https://mlsanity.com/paper/2510.21437/) (2025)
- [HyperVQ: Enabling Hyperprior Entropy Modeling for VQ-Based Generative Image Compression](https://mlsanity.com/paper/2512.07192/) (2025)
- [Deep Light Pollution Removal in Night Cityscape Photographs](https://mlsanity.com/paper/2604.09145/) (2026)
- [LongVQUBench: Benchmarking Long-Term Video Quality Understanding of Vision-Language Models](https://mlsanity.com/paper/2607.01086/) (2026)

## Abstract (arXiv)

Arguably the most common and salient object in daily video communications is the talking head, as encountered in social media, virtual classrooms, teleconferences, news broadcasting, talk shows, etc. When communication bandwidth is limited by network congestions or cost effectiveness, compression artifacts in talking head videos are inevitable. The resulting video quality degradation is highly visible and objectionable due to high acuity of human visual system to faces. To solve this problem, we develop a multi-modality deep convolutional neural network method for restoring face videos that are aggressively compressed. The main innovation is a new DCNN architecture that incorporates known priors of multiple modalities: the video-synchronized speech signal and semantic elements of the compression code stream, including motion vectors, code partition map and quantization parameters. These priors strongly correlate with the latent video and hence they are able to enhance the capability of deep learning to remove compression artifacts. Ample empirical evidences are presented to validate the superior performance of the proposed DCNN method on face videos over the existing state-of-the-art methods.