Research map / Video / Action recognition
MOSCATO: Predicting Multiple Object State Change Through Actions
Parnian Zameni, Yuhan Shen, Ehsan Elhamifar
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 State-Action Interaction (SAI) module (ablation) | MOSP (MSTCN++) · MOSCATO-EgoPER | F1-Max: 49.99 -> 52.83 | Improved |
| add State-Action Interaction (SAI) module (ablation) | MOSP (MSTCN++) · MOSCATO-CMU | F1-Max: 30.55 -> 35.61 | Improved |
| add SAI module (ablation) | MOSP (MSTCN++) · MOSCATO-CMU | F1@0.5: 26.75 -> 29.99 | Improved |
| add SAI module (ablation) | MOSP (MSTCN++) · MOSCATO-EgoPER | mAP: 35.58 -> 35.06 | No clear effect |
| add SAI module (ablation) | MOSP (MSTCN++) · MOSCATO-EGTEA | F1-Max: 12.85 -> 8.04 | Hurt |
| add SAI module (ablation) | MOSP (MSTCN++) · MOSCATO-EGTEA | F1@0.5: 0.0 -> 6.45 | Improved |
| add Stage 3 object state refinement to PLA (ablation) | Pseudo-label pipeline · MOSCATO-CMU pseudo-labels | F1-Max: 7.49 -> 17.96 | Improved |
| use PLA pseudo-labels instead of CLIP for assigning pseudo-labels (ablation) | Pseudo-label pipeline · MOSCATO-CMU pseudo-labels | F1-Max: 12.15 -> 17.96 | Improved |
| add SAI module to TAS branch (ablation) | TAS (MSTCN++) · MOSCATO-EgoPER | Edit: 90.78 -> 91.24 | No clear effect |
| train MOSP on pseudo-labels instead of zero-shot VLM | CLIP / InternVideo2 · MOSCATO-CMU | F1-Max: 12.15 -> 30.55 | Improved |
Closest papers by meaning
- Learning Object State Changes in Videos: An Open-World Perspective (CVPR 2024 · similarity 0.89)
- Look for the Change: Learning Object States and State-Modifying Actions From Untrimmed Web Videos (CVPR 2022 · similarity 0.89)
- PlaySlot: Learning Inverse Latent Dynamics for Controllable Object-Centric Video Prediction and Planning (ICML 2025 · similarity 0.88)
- MolmoMotion: Forecasting Point Trajectories in 3D with Language Instruction (NeurIPS 2026 · similarity 0.86)
- PointZero: 3D Point Track Completion for Learning Transferable 3D Dynamics (NeurIPS 2026 · similarity 0.86)
- Learning to Evolve Scenes: Reasoning about Human Activities with Scene Graphs (NeurIPS 2026 · similarity 0.86)
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.
