mlsanityData to 14 Oct 2026

Exploring Temporally Dynamic Data Augmentation for Video Recognition

arXiv 2206.15015 ↗First posted 30 Jun 2022cs.CVcited by 9

Not on the map. Why?

Abstract (arXiv)

Data augmentation has recently emerged as an essential component of modern training recipes for visual recognition tasks. However, data augmentation for video recognition has been rarely explored despite its effectiveness. Few existing augmentation recipes for video recognition naively extend the image augmentation methods by applying the same operations to the whole video frames. Our main idea is that the magnitude of augmentation operations for each frame needs to be changed over time to capture the real-world video's temporal variations. These variations should be generated as diverse as possible using fewer additional hyper-parameters during training. Through this motivation, we propose a simple yet effective video data augmentation framework, DynaAugment. The magnitude of augmentation operations on each frame is changed by an effective mechanism, Fourier Sampling that parameterizes diverse, smooth, and realistic temporal variations. DynaAugment also includes an extended search space suitable for video for automatic data augmentation methods. DynaAugment experimentally demonstrates that there are additional performance rooms to be improved from static augmentations on diverse video models. Specifically, we show the effectiveness of DynaAugment on various video datasets and tasks: large-scale video recognition (Kinetics-400 and Something-Something-v2), small-scale video recognition (UCF- 101 and HMDB-51), fine-grained video recognition (Diving-48 and FineGym), video action segmentation on Breakfast, video action localization on THUMOS'14, and video object detection on MOT17Det. DynaAugment also enables video models to learn more generalized representation to improve the model robustness on the corrupted videos.

Key facts. Cited by 9 papers in the map (data through 14 Oct 2026). Builds on 69 of its 88 references that are in the map. First posted on arXiv 30 Jun 2022.

Builds on

69 of 88 references.

References · 69

  1. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
  2. 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,364
  3. 2021Swin Transformer: Hierarchical Vision Transformer using Shifted Windowscited by 12,139
  4. 2015Rethinking the Inception Architecture for Computer Visioncited by 5,475
  5. 2017mixup: Beyond Empirical Risk Minimizationcited by 5,183
  6. 2016SGDR: Stochastic Gradient Descent with Warm Restartscited by 5,045
  7. 2020Training data-efficient image transformers & distillation through attentioncited by 4,586
  8. 2022A ConvNet for the 2020scited by 3,579
  9. 2017Quo Vadis, Action Recognition? A New Model and the Kinetics Datasetcited by 3,270
  10. 2019CutMix: Regularization Strategy to Train Strong Classifiers with Localizable Featurescited by 2,884
  11. 2020RAFT: Recurrent All-Pairs Field Transforms for Optical Flowcited by 2,415
  12. 2017Non-local Neural Networkscited by 2,293
38 more
  1. 2019RandAugment: Practical automated data augmentation with a reduced search spacecited by 2,269
  2. 2018SlowFast Networks for Video Recognitioncited by 1,867
  3. 2017Improved Regularization of Convolutional Neural Networks with Cutoutcited by 1,628
  4. 2021Is Space-Time Attention All You Need for Video Understanding?cited by 1,620
  5. 2021ViViT: A Video Vision Transformercited by 1,620
  6. 2017Accurate, Large Minibatch SGD: Training ImageNet in 1 Hourcited by 1,492
  7. 2019SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognitioncited by 1,380
  8. 2017Random Erasing Data Augmentationcited by 1,239
  9. 2019Objects as Pointscited by 1,175
  10. 2017A Closer Look at Spatiotemporal Convolutions for Action Recognitioncited by 1,143
  11. 2017The "something something" video database for learning and evaluating visual common sensecited by 1,103
  12. 2021Video Swin Transformercited by 1,091
  13. 2014Two-Stream Convolutional Networks for Action Recognition in Videoscited by 1,068
  14. 2016Deep Networks with Stochastic Depthcited by 979
  15. 2019AugMix: A Simple Data Processing Method to Improve Robustness and Uncertaintycited by 901
  16. 2021Multiscale Vision Transformerscited by 897
  17. 2018TSM: Temporal Shift Module for Efficient Video Understandingcited by 856
  18. 2019EDA: Easy Data Augmentation Techniques for Boosting Performance on Text Classification Taskscited by 807
  19. 2015Improving Neural Machine Translation Models with Monolingual Datacited by 788
  20. 2020X3D: Expanding Architectures for Efficient Video Recognitioncited by 637
  21. 2017Can Spatiotemporal 3D CNNs Retrace the History of 2D CNNs and ImageNet?cited by 551
  22. 2016MOT16: A Benchmark for Multi-Object Trackingcited by 532
  23. 2017Rethinking Spatiotemporal Feature Learning: Speed-Accuracy Trade-offs in Video Classificationcited by 521
  24. 2019MS-TCN: Multi-Stage Temporal Convolutional Network for Action Segmentationcited by 373
  25. 2020Spatiotemporal Contrastive Video Representation Learningcited by 372
  26. 2021ResNet strikes back: An improved training procedure in timmcited by 368
  27. 2019Video Classification with Channel-Separated Convolutional Networkscited by 290
  28. 2021Video Transformer Networkcited by 272
  29. 2018DropBlock: A regularization method for convolutional networkscited by 262
  30. 2020Puzzle Mix: Exploiting Saliency and Local Statistics for Optimal Mixupcited by 261
  31. 2019G-TAD: Sub-Graph Localization for Temporal Action Detectioncited by 249
  32. 2019Fast AutoAugmentcited by 243
  33. 2020FineGym: A Hierarchical Video Dataset for Fine-grained Action Understandingcited by 234
  34. 2021TrivialAugment: Tuning-free Yet State-of-the-Art Data Augmentationcited by 227
  35. 2016FractalNet: Ultra-Deep Neural Networks without Residualscited by 216
  36. 2017Hide-and-Seek: Forcing a Network to be Meticulous for Weakly-supervised Object and Action Localizationcited by 207
  37. 2020MixText: Linguistically-Informed Interpolation of Hidden Space for Semi-Supervised Text Classificationcited by 187
  38. 2022UniFormer: Unified Transformer for Efficient Spatiotemporal Representation Learningcited by 184

50 most cited of 69 All 69 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2023: 2; 2024: 5; 2025: 1; 2026: 1.

9 papers.

  1. 2024Universal Time-Series Representation Learning: A Surveycited by 30
  2. 2024Benchmarking the Robustness of Temporal Action Detection Models Against Temporal Corruptionscited by 12
  3. 2024Flow Snapshot Neurons in Action: Deep Neural Networks Generalize to Biological Motion Perceptioncited by 5
  4. 2024Multi-Scale Contrastive Learning for Video Temporal Groundingcited by 5
  5. 2025Vision-Based Anti Unmanned Aerial Technology: Opportunities and Challengescited by 4
  6. 2023Selective Volume Mixup for Video Action Recognitioncited by 2
  7. 2023Semantic Equivariant Mixupcited by 0
  8. 2024SDI-Paste: Synthetic Dynamic Instance Copy-Paste for Video Instance Segmentationcited by 0
  9. 2026Robust Promptable Video Object Segmentationcited by 0