Research map / Video / Action recognition
Forecasting Action Through Contact Representations From First Person Video
Eadom Dessalene, Chinmaya Devaraj, Michael Maynord, Cornelia M. Fermuller, Yiannis Aloimonos
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 the GCN stream to the CSN stream (ablation) | CSN stream · EPIC val, anticipation tau_a=1s | accuracy (%): 15.50 -> 19.20 | Improved |
| add the GCN stream to the CSN stream (ablation) | CSN stream · EPIC val, anticipation tau_a=5s | accuracy (%): 6.49 -> 9.44 | Improved |
| add the GCN stream to the CSN stream (ablation) | CSN stream · EPIC val, action prediction p=90% | accuracy (%): 31.30 -> 31.42 | Improved |
| add the CSN stream to the GCN stream (ablation) | GCN stream · EPIC val, anticipation tau_a=1s | accuracy (%): 12.81 -> 19.20 | Improved |
| use Ego-OMG instead of RULSTM at long observation | RULSTM · EPIC val, action prediction p=75% and 90% | accuracy (%): 31.19 vs 33.09; 31.42 vs 34.07 | Hurt |
| use Ego-OMG instead of RULSTM at long anticipation time | RULSTM · EPIC val, anticipation tau_a=5s | accuracy (%): 6.98 -> 9.44 | Improved |
| use Ego-OMG instead of previous best in unseen kitchens | Liu et al. · EPIC Kitchens test S2 | top-1 accuracy (%): 9.94 -> 11.80 | Improved |
| use Ego-OMG instead of previous best in seen kitchens | Camp. et al. · EPIC Kitchens test S1 | top-1 accuracy (%): 15.67 -> 16.02 | Improved |
| add Contact Anticipation Maps to Next Active Object network (ablation) | Ours w/o CAM · Next Active Object localization | Jaccard: 0.169 -> 0.194 | Improved |
| model left and right hands separately instead of jointly (ablation) | Joint hands · EPIC val, anticipation tau_a=1s | top-1 accuracy (%): 12.12 -> 12.81 | Improved |
| model left and right hands separately instead of jointly (ablation) | Joint hands · EPIC val, action prediction p=25% | top-1 accuracy (%): 14.63 -> 15.44 | Improved |
| add Next Active Object predictions to Active Object only (ablation) | AO only · EPIC val, anticipation tau_a=1s | top-1 accuracy (%): 10.91 -> 12.81 | Improved |
Closest papers by meaning
- Forecasting Human-Object Interaction: Joint Prediction of Motor Attention and Actions in First Person Video (ECCV 2020 · similarity 0.92)
- Joint Hand Motion and Interaction Hotspots Prediction From Egocentric Videos (CVPR 2022 · similarity 0.91)
- Intention-Guided Cognitive Reasoning for Egocentric Long-Term Action Anticipation (AAAI 2026 · similarity 0.91)
- Summarize the Past to Predict the Future: Natural Language Descriptions of Context Boost Multimodal Object Interaction Anticipation (CVPR 2024 · similarity 0.89)
- Modeling Fine-Grained Hand-Object Dynamics for Egocentric Video Representation Learning (ICLR 2025 · similarity 0.88)
- How Do I Do That? Synthesizing 3D Hand Motion and Contacts for Everyday Interactions (CVPR 2025 · similarity 0.87)
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.
