Research map / Video / Action recognition
Weakly Supervised Learning of Heterogeneous Concepts in Videos
Sohil Shah, Kuldeep Kulkarni, Arijit Biswas, Ankit Gandhi, Om Deshmukh, Larry S. Davis
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 integrative heterogeneous-concept modeling (ablation) | WS-SIBP · A2D (localization) | average IoU (%): 31.1 -> 31.55 | Improved |
| add location constraints (ablation) | WS-SIBP · A2D (localization) | average IoU (%): 31.1 -> 31.69 | Improved |
| add location constraints (ablation) | WS-SIIBP · A2D (localization) | average IoU (%): 31.55 -> 34.38 | Improved |
| add integrative heterogeneous-concept modeling (ablation) | WSC-SIBP · A2D (localization) | average IoU (%): 31.69 -> 34.38 | Improved |
| use WSC-SIIBP instead of next best baseline | localization · A2D (localization) | average IoU (%): 31.55 -> 34.38 | Improved |
| provide no test video labels (free annotation) | WSC-SIIBP · A2D (test) | object mAP (%): 94.77 -> 76.62 | Hurt |
| provide no test video labels (free annotation) | WSC-SIIBP · A2D (test) | action mAP (%): 90.68 -> 64.77 | Hurt |
| use fully supervised GT-SVM instead of weak supervision | WSC-SIIBP · A2D (test, video labels) | object mAP (%): 94.77 -> 98.20 | Improved |
| use WSC-SIIBP instead of WS-SIIBP on 120 s segments | person classification · Casablanca (120 s) | person classification accuracy: +4% | Improved |
| use WSC-SIIBP instead of most competitive baseline | pairwise classification · Casablanca | pairwise classification accuracy: +8.1% | Improved |
| add constraints to WS-DC (ablation) | WS-DC · Casablanca | classification accuracy | Improved |
Closest papers by meaning
- Weakly-Supervised Action Localization With Background Modeling (ICCV 2019 · similarity 0.85)
- Weakly-Supervised Spatio-Temporal Video Grounding with Variational Cross-Modal Alignment (ECCV 2024 · similarity 0.85)
- Weakly-supervised Temporal Action Localization by Uncertainty Modeling (AAAI 2021 · similarity 0.85)
- Local-Global Multi-Modal Distillation for Weakly-Supervised Temporal Video Grounding (AAAI 2024 · similarity 0.84)
- Contextual Self-paced Learning for Weakly Supervised Spatio-Temporal Video Grounding (ICLR 2025 · similarity 0.84)
- Omnipotent Distillation with LLMs for Weakly-Supervised Natural Language Video Localization: When Divergence Meets Consistency (AAAI 2024 · similarity 0.83)
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.
