mlsanityData to 14 Oct 2026

ViewBridge: Curriculum Knowledge Distillation for Activity View-Invariance Under Extreme Viewpoint Changes

arXiv 2504.05451 ↗First posted 7 Apr 2025cs.CVcited by 0

On the map: Knowledge distillation → Detection and dense prediction distillation

Abstract (arXiv)

Traditional methods for view-invariant learning rely on controlled multi-view training data with minimal scene clutter. However, they struggle with in-the-wild videos that exhibit extreme viewpoint differences and share little visual content. We introduce ViewBridge, a framework for learning rich video representations in the presence of severe view-occlusions. We introduce a knowledge distillation objective that preserves action-centric semantics, together with a novel curriculum learning procedure that pairs incrementally more challenging views over time, thereby allowing smooth adaptation to extreme viewpoint differences. To sort training video segments for the proposed curriculum, we define a geometry-based metric that reflects their likely occlusion level. While training leverages multi-view data, at inference time, the input is an uncalibrated, single-viewpoint video. Evaluating our approach on two tasks -- temporal keystep grounding and fine-grained keystep recognition -- we outperform SOTA approaches across three datasets (Ego-Exo4D, LEMMA, EPFL-Smart-Kitchen-30). Project page: https://vision.cs.utexas.edu/projects/learning_view_distill/ .

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Detection and dense prediction distillation, one of 12 directions in Knowledge distillation. Builds on 33 of its 57 references that are in the map. First posted on arXiv 7 Apr 2025.

Builds on

33 of 57 references.

References · 33

  1. 2021Is Space-Time Attention All You Need for Video Understanding?cited by 1,620
  2. 2015Unsupervised Visual Representation Learning by Context Predictioncited by 1,066
  3. 2019HowTo100M: Learning a Text-Video Embedding by Watching Hundred Million Narrated Video Clipscited by 950
  4. 2016NTU RGB+D: A Large Scale Dataset for 3D Human Activity Analysiscited by 686
  5. 2017Towards Automatic Learning of Procedures from Web Instructional Videoscited by 619
  6. 2019NTU RGB+D 120: A Large-Scale Benchmark for 3D Human Activity Understandingcited by 533
  7. 2022Assembly101: A Large-Scale Multi-View Video Dataset for Understanding Procedural Activitiescited by 287
  8. 2019Cross-task weakly supervised learning from instructional videoscited by 199
  9. 2020Dense Regression Network for Video Groundingcited by 176
  10. 2023EgoVLPv2: Egocentric Video-Language Pre-training with Fusion in the Backbonecited by 130
  11. 2017View Adaptive Recurrent Neural Networks for High Performance Human Action Recognition from Skeleton Datacited by 105
  12. 2021Ego-Exo: Transferring Visual Representations from Third-person to First-person Videoscited by 99
21 more
  1. 2022Temporal Alignment Networks for Long-term Videocited by 84
  2. 2024Reloc3r: Large-Scale Training of Relative Camera Pose Regression for Generalizable, Fast, and Accurate Visual Localizationcited by 82
  3. 2022My View is the Best View: Procedure Learning from Egocentric Videoscited by 81
  4. 2019View-Invariant Probabilistic Embedding for Human Posecited by 47
  5. 2020LEMMA: A Multi-view Dataset for Learning Multi-agent Multi-task Activitiescited by 47
  6. 2023Learning Fine-grained View-Invariant Representations from Unpaired Ego-Exo Videos via Temporal Alignmentcited by 46
  7. 2022Temporal Sentence Grounding in Videos: A Survey and Future Directionscited by 37
  8. 2023Video-Mined Task Graphs for Keystep Recognition in Instructional Videoscited by 35
  9. 2018Unsupervised Learning of View-invariant Action Representationscited by 32
  10. 2023On the Benefits of 3D Pose and Tracking for Human Action Recognitioncited by 32
  11. 2024Put Myself in Your Shoes: Lifting the Egocentric Perspective from Exocentric Videoscited by 32
  12. 2023Learning to Ground Instructional Articles in Videos through Narrationscited by 28
  13. 2023Localizing Moments in Long Video Via Multimodal Guidancecited by 23
  14. 2020Comprehensive Instructional Video Analysis: The COIN Dataset and Performance Evaluationcited by 20
  15. 2021ViewCLR: Learning Self-supervised Video Representation for Unseen Viewpointscited by 20
  16. 2021Recognizing Actions in Videos from Unseen Viewpointscited by 18
  17. 2019What I See Is What You See: Joint Attention Learning for First and Third Person Video Co-analysiscited by 17
  18. 2023Learning Distortion Invariant Representation for Image Restoration from A Causality Perspectivecited by 17
  19. 2024Which Viewpoint Shows it Best? Language for Weakly Supervising View Selection in Multi-view Instructional Videoscited by 6
  20. 2021Unsupervised View-Invariant Human Posture Representationcited by 3
  21. 2023Learning Temporal Sentence Grounding From Narrated EgoVideoscited by 3

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026Improving Viewpoint-Invariance and Temporal Consistency for Action Detection
  2. 2026Deep Thought Alignment: Trajectory-Level Latent Distillation for Video Reasoning
  3. 2026Single-Teacher View Augmentation: Enhancing Knowledge Distillation with Student-Guided Perturbations
  4. 2025DUNE: Distilling a Universal Encoder from Heterogeneous 2D and 3D Teachers
  5. 2025GLaD: Geometric Latent Distillation for Vision-Language-Action Models
  6. 2024Omniview-Tuning: Boosting Viewpoint Invariance of Vision-Language Pre-training Models
  7. 2026DVTR-KD: Distilling Vision-Language Models via Intermediate Visual Token Reconstruction
  8. 2026Chain-of-Glimpse: Search-Guided Progressive Object-Grounded Reasoning for Video Understanding