Research map / Video / Action recognition
PivoTAL: Prior-Driven Supervision for Weakly-Supervised Temporal Action Localization
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Add a Gaussian prior for actionness prediction (ablation) | Base WTAL · THUMOS-14 | average mAP: 35.8 -> 38.4 | Helped |
| Add the scene prior to the Gaussian-prior model (ablation) | Base WTAL · THUMOS-14 | average mAP: 38.4 -> 39.7 | Helped |
| Train the localization head with hard pseudo-action snippets (ablation) | Base WTAL with Gaussian and scene priors · THUMOS-14 | average mAP: 39.7 -> 39.0 | Hurt |
| Use soft pseudo-action snippets and per-class confidence normalization instead of hard snippets (ablation) | PivoTAL localization head · THUMOS-14 | average mAP: 39.0 -> 46.6 | Helped |
| Add MIL loss to localization-head training (ablation) | PivoTAL · THUMOS-14 | average mAP: 46.6 -> 49.6 | Helped |
| Use clip-specific local Gaussian masks rather than a single global mask (ablation) | PivoTAL · THUMOS-14 | average mAP: 45.2 -> 49.6 | Helped |
| Use clip-specific local Gaussian masks rather than five global masks (ablation) | PivoTAL · THUMOS-14 | average mAP: 41.3 -> 49.6 | Helped |
| Use clip-specific local Gaussian masks rather than ten global masks (ablation) | PivoTAL · THUMOS-14 | average mAP: 44.0 -> 49.6 | Helped |
| Select each clip's corresponding value from its local Gaussian mask rather than averaging mask contributions (ablation) | PivoTAL · THUMOS-14 | average mAP: 43.0 -> 49.6 | Helped |
| Use a Gaussian-prior actionness head rather than a second learning-based head (ablation) | PivoTAL · THUMOS-14 | average mAP: 48.2 -> 49.6 | Helped |
| Average the learning-based and Gaussian-based actionness scores rather than multiplying them (ablation) | PivoTAL · THUMOS-14 | average mAP: 45.5 -> 49.6 | Helped |
| Average the learning-based and Gaussian-based actionness scores rather than taking their maximum (ablation) | PivoTAL · THUMOS-14 | average mAP: 48.0 -> 49.6 | Helped |
Closest papers by meaning
- Bridge the Gap: From Weak to Full Supervision for Temporal Action Localization with PseudoFormer (CVPR 2025 · similarity 0.93)
- ASM-Loc: Action-Aware Segment Modeling for Weakly-Supervised Temporal Action Localization (CVPR 2022 · similarity 0.93)
- Weakly-Supervised Temporal Action Localization by Inferring Salient Snippet-Feature (AAAI 2024 · similarity 0.92)
- Similar Modality Enhancement and Action Consistency Learning for Weakly Supervised Temporal Action Localization (AAAI 2025 · similarity 0.92)
- A flexible model for training action localization with varying levels of supervision (NeurIPS 2018 · similarity 0.92)
- Weakly Supervised Temporal Action Localization Through Learning Explicit Subspaces for Action and Context (AAAI 2021 · similarity 0.92)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.