Research map / Video / Action recognition
Object Priors for Classifying and Localizing Unseen Actions
Pascal Mettes, William E. Thong, Cees G. M. Snoek
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 local object detections to person detections (ablation) | spatial object priors · UCF Sports (unseen) | classification accuracy: 8.5 -> 27.7 | Improved |
| add spatial preposition prior (ablation) | spatial object priors (5 objects) · UCF Sports (unseen) | classification accuracy: 27.7 -> 29.8 | Improved |
| add spatial preposition prior (ablation) | spatial object priors (5 objects) · UCF Sports (unseen) | localization mAP@0.5: 24.4 -> 27.0 | Improved |
| increase local objects from 5 to 10 (ablation) | spatial object priors with prepositions · UCF Sports (unseen) | localization mAP@0.5: 27.0 -> 22.8 | Hurt |
| add Dutch to English in semantic matching (ablation) | semantic ambiguity prior · UCF-101 (50 classes) | accuracy: 43.3 -> 45.8 | Improved |
| use all six languages instead of English and Dutch (ablation) | semantic ambiguity prior · UCF-101 (50 classes) | accuracy: 45.8 -> 43.2 | Hurt |
| use a non-English language alone instead of English (ablation) | semantic matching · UCF-101 (50 classes) | accuracy: 43.3 -> 31.2 (German) | Hurt |
| add object-based discrimination (ablation) | semantic matching · UCF-101 (50 classes) | accuracy: 43.3 -> 44.7 | Improved |
| add action-based discrimination (ablation) | semantic matching · UCF-101 (101 classes) | accuracy: 33.3 -> 34.3 | Improved |
| basic-level object weighting instead of uniform (ablation) | object naming prior · UCF-101 (50 classes) | accuracy: 43.3 -> 43.9 | No clear effect |
| weight specific objects only (ablation) | object naming prior · UCF-101 (50 classes) | accuracy: 43.3 -> 7.6 | Hurt |
| add discrimination prior on top of ambiguity prior (ablation) | semantic priors · Kinetics-400 (400 actions) | accuracy: 6.3 -> 6.1 | No clear effect |
Closest papers by meaning
- Spatial-Aware Object Embeddings for Zero-Shot Localization and Classification of Actions (ICCV 2017 · similarity 0.93)
- Unsupervised Open-Vocabulary Object Localization in Videos (ICCV 2023 · similarity 0.90)
- AVA: A Video Dataset of Spatio-Temporally Localized Atomic Visual Actions (CVPR 2018 · similarity 0.89)
- Universal Prototype Transport for Zero-Shot Action Recognition and Localization (IJCV 2023 · similarity 0.89)
- Learning Object State Changes in Videos: An Open-World Perspective (CVPR 2024 · similarity 0.88)
- VideoGEM: Training-free Action Grounding in Videos (CVPR 2025 · similarity 0.88)
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.
