Summary of Finepseudo: Improving Pseudo-labelling Through Temporal-alignablity For Semi-supervised Fine-grained Action Recognition, by Ishan Rajendrakumar Dave et al.
FinePseudo: Improving Pseudo-Labelling through Temporal-Alignablity for Semi-Supervised Fine-Grained Action Recognition
by Ishan Rajendrakumar Dave, Mamshad Nayeem Rizve, Mubarak Shah
First submitted to arxiv on: 2 Sep 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper presents a novel approach to semi-supervised fine-grained action recognition, a crucial task for applications such as sports analytics, user interactions in AR/VR, and surgical videos. Existing methods primarily focus on coarse-grained action recognition, neglecting the complexity of fine-grained actions. The authors propose an Alignability-Verification-based Metric learning technique to effectively discriminate between fine-grained action pairs using dynamic time warping (DTW) distances. They introduce a learnable alignability score that refines pseudo-labels for the primary video encoder, leading to improved performance on four fine-grained datasets and two coarse-grained datasets. The collaborative pseudo-labeling-based framework, FinePseudo, demonstrates robustness in open-world semi-supervised setups. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper explores a new way to recognize actions with high precision. Currently, action recognition mostly focuses on big movements, but this paper looks at small actions like gestures or specific movements in sports. This is important because these small actions can tell us more about what’s happening in a scene. The authors develop a method that uses alignment distances to compare these small actions and refine their predictions. Their approach, called FinePseudo, performs better than existing methods on several action recognition datasets. |
Keywords
» Artificial intelligence » Alignment » Encoder » Precision » Semi supervised
