mlsanityData to 14 Oct 2026

Minimalistic Video Saliency Prediction via Efficient Decoder & Spatio Temporal Action Cues

arXiv 2502.00397 ↗First posted 1 Feb 2025cs.CVcited by 4

Not on the map. Why?

Abstract (arXiv)

This paper introduces ViNet-S, a 36MB model based on the ViNet architecture with a U-Net design, featuring a lightweight decoder that significantly reduces model size and parameters without compromising performance. Additionally, ViNet-A (148MB) incorporates spatio-temporal action localization (STAL) features, differing from traditional video saliency models that use action classification backbones. Our studies show that an ensemble of ViNet-S and ViNet-A, by averaging predicted saliency maps, achieves state-of-the-art performance on three visual-only and six audio-visual saliency datasets, outperforming transformer-based models in both parameter efficiency and real-time performance, with ViNet-S reaching over 1000fps.

Key facts. Cited by 4 papers in the map (data through 14 Oct 2026). Builds on 25 of its 41 references that are in the map. First posted on arXiv 1 Feb 2025.

Builds on

25 of 41 references.

References · 25

  1. 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,713
  2. 2017The Kinetics Human Action Video Datasetcited by 1,909
  3. 2018SlowFast Networks for Video Recognitioncited by 1,867
  4. 2017ShuffleNet: An Extremely Efficient Convolutional Neural Network for Mobile Devicescited by 1,200
  5. 2017Rethinking Spatiotemporal Feature Learning: Speed-Accuracy Trade-offs in Video Classificationcited by 521
  6. 2017AVA: A Video Dataset of Spatio-temporally Localized Atomic Visual Actionscited by 449
  7. 2016What do different evaluation metrics tell us about saliency models?cited by 185
  8. 2020Actor-Context-Actor Relation Network for Spatio-Temporal Action Localizationcited by 91
  9. 2020Unified Image and Video Saliency Modelingcited by 75
  10. 2019TASED-Net: Temporally-Aggregating Spatial Encoder-Decoder Network for Video Saliency Detectioncited by 59
  11. 2020ViNet: Pushing the limits of Visual Modality for Audio-Visual Saliency Predictioncited by 43
  12. 2020STAViS: Spatio-Temporal AudioVisual Saliency Networkcited by 40
13 more
  1. 2013Actions in the Eye: Dynamic Gaze Datasets and Learnt Saliency Models for Visual Recognitioncited by 37
  2. 2016Deep Roots: Improving CNN Efficiency with Hierarchical Filter Groupscited by 32
  3. 2021Temporal-Spatial Feature Pyramid for Video Saliency Detectioncited by 22
  4. 2021Spatio-Temporal Self-Attention Network for Video Saliency Predictioncited by 19
  5. 2023CASP-Net: Rethinking Video Saliency Prediction from an Audio-VisualConsistency Perceptual Perspectivecited by 15
  6. 2024DiffSal: Joint Audio and Video Learning for Diffusion Saliency Predictioncited by 13
  7. 2020GAZED- Gaze-guided Cinematic Editing of Wide-Angle Monocular Video Recordingscited by 11
  8. 2023TinyHD: Efficient Video Saliency Prediction with Heterogeneous Decoders using Hierarchical Maps Distillationcited by 9
  9. 2024Transformer-based Video Saliency Prediction with High Temporal Dimension Decodingcited by 7
  10. 2017Facial Expression Recognition using Visual Saliency and Deep Learningcited by 6
  11. 2021Learning to Predict Salient Faces: A Novel Visual-Audio Saliency Modelcited by 6
  12. 2019SalGaze: Personalizing Gaze Estimation Using Visual Saliencycited by 4
  13. 2021A Gated Fusion Network for Dynamic Saliency Predictioncited by 3

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 3; 2026: 1.

4 papers.

  1. 2025DTFSal: Audio-Visual Dynamic Token Fusion for Video Saliency Predictioncited by 1
  2. 2025DriverGaze360: OmniDirectional Driver Attention with Object-Level Guidancecited by 1
  3. 2025Simplifying Knowledge Transfer in Pretrained Modelscited by 0
  4. 2026ViASNet: A Video Ad Saliency Network for Predicting Dynamic Saliency and Viewer Engagementcited by 0