mlsanityData to 14 Oct 2026

PECoP: Parameter Efficient Continual Pretraining for Action Quality Assessment

arXiv 2311.07603 ↗First posted 11 Nov 2023cs.CVcited by 19

Not on the map. Why?

Abstract (arXiv)

The limited availability of labelled data in Action Quality Assessment (AQA), has forced previous works to fine-tune their models pretrained on large-scale domain-general datasets. This common approach results in weak generalisation, particularly when there is a significant domain shift. We propose a novel, parameter efficient, continual pretraining framework, PECoP, to reduce such domain shift via an additional pretraining stage. In PECoP, we introduce 3D-Adapters, inserted into the pretrained model, to learn spatiotemporal, in-domain information via self-supervised learning where only the adapter modules' parameters are updated. We demonstrate PECoP's ability to enhance the performance of recent state-of-the-art methods (MUSDL, CoRe, and TSA) applied to AQA, leading to considerable improvements on benchmark datasets, JIGSAWS (uparrow6.0%), MTL-AQA (uparrow0.99%), and FineDiving (uparrow2.54%). We also present a new Parkinson's Disease dataset, PD4T, of real patients performing four various actions, where we surpass (uparrow3.56%) the state-of-the-art in comparison. Our code, pretrained models, and the PD4T dataset are available at https://github.com/Plrbear/PECoP.

Key facts. Cited by 19 papers in the map (data through 14 Oct 2026). Builds on 40 of its 54 references that are in the map. First posted on arXiv 11 Nov 2023.

Builds on

40 of 54 references.

References · 40

  1. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
  2. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,380
  3. 2020A Simple Framework for Contrastive Learning of Visual Representationscited by 12,049
  4. 2019Momentum Contrast for Unsupervised Visual Representation Learningcited by 7,513
  5. 2019Parameter-Efficient Transfer Learning for NLPcited by 4,236
  6. 2017Quo Vadis, Action Recognition? A New Model and the Kinetics Datasetcited by 3,270
  7. 2019BioBERT: a pre-trained biomedical language representation model for biomedical text miningcited by 1,947
  8. 2017The Kinetics Human Action Video Datasetcited by 1,909
  9. 2020Don't Stop Pretraining: Adapt Language Models to Domains and Taskscited by 1,565
  10. 2021Towards a Unified View of Parameter-Efficient Transfer Learningcited by 872
  11. 2022AdaptFormer: Adapting Vision Transformers for Scalable Visual Recognitioncited by 719
  12. 2020AdapterHub: A Framework for Adapting Transformerscited by 444
28 more
  1. 2017Learning Spatio-Temporal Features with 3D Residual Networks for Action Recognitioncited by 178
  2. 2020SpeedNet: Learning the Speediness in Videoscited by 171
  3. 2021On the Effectiveness of Adapter-based Tuning for Pretrained Language Model Adaptationcited by 147
  4. 2020Self-supervised Video Representation Learning by Pace Predictioncited by 139
  5. 2022FineDiving: A Fine-grained Dataset for Procedure-aware Action Quality Assessmentcited by 107
  6. 2020Training BatchNorm and Only BatchNorm: On the Expressive Power of Random Features in CNNscited by 104
  7. 2019What and How Well You Performed? A Multitask Learning Approach to Action Quality Assessmentcited by 100
  8. 2020RSPNet: Relative Speed Perception for Unsupervised Video Representation Learningcited by 83
  9. 2019Oops! Predicting Unintentional Action in Videocited by 82
  10. 2021Self-Supervised Pretraining Improves Self-Supervised Pretrainingcited by 82
  11. 2020Uncertainty-aware Score Distribution Learning for Action Quality Assessmentcited by 80
  12. 2018Action Quality Assessment Across Multiple Actionscited by 79
  13. 2021Group-aware Contrastive Regression for Action Quality Assessmentcited by 75
  14. 2018Learning to score the figure skating sports videoscited by 73
  15. 2022Conv-Adapter: Exploring Parameter Efficient Transfer Learning for ConvNetscited by 62
  16. 2021Towards Unified Surgical Skill Assessmentcited by 49
  17. 2022Robust and Efficient Medical Imaging with Self-Supervisioncited by 40
  18. 2022TSA-Net: Tube Self-Attention Network for Action Quality Assessmentcited by 36
  19. 2022Action Quality Assessment with Temporal Parsing Transformercited by 31
  20. 2022Domain Knowledge-Informed Self-Supervised Representations for Workout Form Assessmentcited by 25
  21. 2021Piano Skills Assessmentcited by 16
  22. 2022C3: Continued Pretraining with Contrastive Weak Supervision for Cross Language Ad-Hoc Retrievalcited by 14
  23. 2022Surgical Skill Assessment via Video Semantic Aggregationcited by 13
  24. 2021Domain-Adaptive Pretraining Methods for Dialogue Understandingcited by 11
  25. 2022GaitForeMer: Self-Supervised Pre-Training of Transformers via Human Motion Forecasting for Few-Shot Gait Impairment Severity Estimationcited by 11
  26. 2020Exploring Motion Boundaries in an End-to-End Network for Vision-based Parkinson's Severity Assessmentcited by 4
  27. 2021Improving Action Quality Assessment using Weighted Aggregationcited by 4
  28. 2021Auxiliary Learning for Self-Supervised Video Representation via Similarity-based Knowledge Distillationcited by 3

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2023: 1; 2024: 4; 2025: 10; 2026: 4.

19 papers.

Sort:

By area:

  1. 2024GAIA: Rethinking Action Quality Assessment for AI-Generated Videoscited by 30
  2. 2025Action Quality Assessment via Hierarchical Pose-guided Multi-stage Contrastive Regressioncited by 21
  3. 2024A Comprehensive Survey of Action Quality Assessment: Method and Benchmarkcited by 20
  4. 2024Hierarchical NeuroSymbolic Approach for Comprehensive and Explainable Action Quality Assessmentcited by 14
  5. 2025PHI: Bridging Domain Shift in Long-Term Action Quality Assessment via Progressive Hierarchical Instructioncited by 10
  6. 2025GAITGen: Disentangled Motion-Pathology Impaired Gait Generative Model -- Bringing Motion Generation to the Clinical Domaincited by 7
  7. 2024Snuffy: Efficient Whole Slide Image Classifiercited by 6
  8. 2025CARE-PD: A Multi-Site Anonymized Clinical Dataset for Parkinson's Disease Gait Assessmentcited by 6
  9. 2023Collaborative Weakly Supervised Video Correlation Learning for Procedure-Aware Instructional Video Analysiscited by 4
  10. 2025A Methodological and Structural Review of Parkinsons Disease Detection Across Diverse Data Modalitiescited by 4
  11. 2025Co-STAR: Collaborative Curriculum Self-Training with Adaptive Regularization for Source-Free Video Domain Adaptationcited by 2
  12. 2025Learning Long-Range Action Representation by Two-Stream Mamba Pyramid Network for Figure Skating Assessmentcited by 2
7 more
  1. 2025EHWGesture -- A dataset for multimodal understanding of clinical gesturescited by 2
  2. 2026Aggregate, Don't Adapt: Subject-Level Posterior Aggregation and Transductive Calibration for Cross-Site Parkinsonian Gait Severitycited by 1
  3. 2025LiLAW: Lightweight Learnable Adaptive Weighting to Learn Sample Difficulty & Improve Noisy Trainingcited by 0
  4. 2025Skeleton-Snippet Contrastive Learning with Multiscale Feature Fusion for Action Localizationcited by 0
  5. 2026SurgFusion-Net: Diversified Adaptive Multimodal Fusion Network for Surgical Skill Assessmentcited by 0
  6. 2026AgriField-40K: Adapting Vision Models to Agriculture With Efficient Continual Pretrainingcited by 0
  7. 2026SynthGait-19K: A Physically Grounded Synthetic Video Dataset for Gait Parameter Estimationcited by 0