# Learning to Predict Salient Faces: A Novel Visual-Audio Saliency Model

arXiv: https://arxiv.org/abs/2103.15438 · HTML: https://mlsanity.com/paper/2103.15438/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Builds on 15 of its 37 references that are in the map.
- First posted on arXiv 29 Mar 2021.

## Builds on (most cited first, 15 of 15)

- [Very Deep Convolutional Networks for Large-Scale Image Recognition](https://mlsanity.com/paper/1409.1556/) (2014)
- [Convolutional LSTM Network: A Machine Learning Approach for Precipitation Nowcasting](https://mlsanity.com/paper/1506.04214/) (2015)
- [FlowNet: Learning Optical Flow with Convolutional Networks](https://mlsanity.com/paper/1504.06852/) (2015)
- [Joint Face Detection and Alignment using Multi-task Cascaded Convolutional Networks](https://mlsanity.com/paper/1604.02878/) (2016)
- [CNN Architectures for Large-Scale Audio Classification](https://mlsanity.com/paper/1609.09430/) (2016)
- [What do different evaluation metrics tell us about saliency models?](https://mlsanity.com/paper/1604.03605/) (2016)
- [Predicting Human Eye Fixations via an LSTM-based Saliency Attentive Model](https://mlsanity.com/paper/1611.09571/) (2016)
- [Deep Visual Attention Prediction](https://mlsanity.com/paper/1705.02544/) (2017)
- [Revisiting Video Saliency: A Large-scale Benchmark and a New Model](https://mlsanity.com/paper/1801.07424/) (2018)
- [SalGAN: Visual Saliency Prediction with Generative Adversarial Networks](https://mlsanity.com/paper/1701.01081/) (2017)
- [TASED-Net: Temporally-Aggregating Spatial Encoder-Decoder Network for Video Saliency Detection](https://mlsanity.com/paper/1908.05786/) (2019)
- [Predicting Head Movement in Panoramic Video: A Deep Reinforcement Learning Approach](https://mlsanity.com/paper/1710.10755/) (2017)
- [DAVE: A Deep Audio-Visual Embedding for Dynamic Saliency Prediction](https://mlsanity.com/paper/1905.10693/) (2019)
- [Bridge the Gap Between VQA and Human Behavior on Omnidirectional Video: A Large-Scale Dataset and a Deep Learning Model](https://mlsanity.com/paper/1807.10990/) (2018)
- [Predicting Video Saliency with Object-to-Motion CNN and Two-layer Convolutional LSTM](https://mlsanity.com/paper/1709.06316/) (2017)

## Built on by (most cited first, 6 of 6)

- [CaRDiff: Video Salient Object Ranking Chain of Thought Reasoning for Saliency Prediction with Diffusion](https://mlsanity.com/paper/2408.12009/) (2024)
- [Minimalistic Video Saliency Prediction via Efficient Decoder & Spatio Temporal Action Cues](https://mlsanity.com/paper/2502.00397/) (2025)
- [Joint Learning of Visual-Audio Saliency Prediction and Sound Source Localization on Multi-face Videos](https://mlsanity.com/paper/2111.08567/) (2021)
- [Unified Dynamic Scanpath Predictors Outperform Individually Trained Neural Models](https://mlsanity.com/paper/2405.02929/) (2024)
- [DTFSal: Audio-Visual Dynamic Token Fusion for Video Saliency Prediction](https://mlsanity.com/paper/2504.10070/) (2025)
- [ViASNet: A Video Ad Saliency Network for Predicting Dynamic Saliency and Viewer Engagement](https://mlsanity.com/paper/2605.29302/) (2026)

## Abstract (arXiv)

Recently, video streams have occupied a large proportion of Internet traffic, most of which contain human faces. Hence, it is necessary to predict saliency on multiple-face videos, which can provide attention cues for many content based applications. However, most of multiple-face saliency prediction works only consider visual information and ignore audio, which is not consistent with the naturalistic scenarios. Several behavioral studies have established that sound influences human attention, especially during the speech turn-taking in multiple-face videos. In this paper, we thoroughly investigate such influences by establishing a large-scale eye-tracking database of Multiple-face Video in Visual-Audio condition (MVVA). Inspired by the findings of our investigation, we propose a novel multi-modal video saliency model consisting of three branches: visual, audio and face. The visual branch takes the RGB frames as the input and encodes them into visual feature maps. The audio and face branches encode the audio signal and multiple cropped faces, respectively. A fusion module is introduced to integrate the information from three modalities, and to generate the final saliency map. Experimental results show that the proposed method outperforms 11 state-of-the-art saliency prediction works. It performs closer to human multi-modal attention.