Research map / Video / Video-language understanding
Hier-EgoPack: Hierarchical Egocentric Video Understanding With Diverse Task Perspectives
Simone Alberto Peirone, Francesca Pistilli, Antonio Alliegro, Tatiana Tommasi, Giuseppe Averta
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 |
|---|---|---|---|
| use Hier-EgoPack cross-task prototypes instead of single-task training | Single Task · Ego4D MQ (val) | mAP: 20.2 -> 21.0 | Improved |
| use Hier-EgoPack for AR nouns instead of single-task training | Single Task · Ego4D AR | noun top-1 accuracy: 33.50 -> 34.65 | Improved |
| train all tasks jointly with hard parameter sharing (MTL) | Single Task · Ego4D MQ (val) | mAP: 20.2 -> 18.5 | Hurt |
| train all tasks jointly with hard parameter sharing (MTL) | Single Task · Ego4D OSCC | accuracy: 75.22 -> 74.79 | Hurt |
| use task-translation instead of single-task on AR verbs | Single Task · Ego4D AR | verb top-1 accuracy: 26.93 -> 26.10 | Hurt |
| use Hier-EgoPack instead of EgoPack with Omnivore features | EgoPack · Ego4D AR (Omnivore) | verb top-1 accuracy: 25.10 -> 25.33 | No clear effect |
| add past/future sign s_ij to TDGC (ablation) | TDGC · Ego4D MQ (val) | avg mAP: 15.12 -> 20.21 | Improved |
| add relative-distance weights w_ij to TDGC (ablation) | TDGC · Ego4D MQ (val) | avg mAP: 19.52 -> 20.21 | Improved |
| use TDGC instead of SGCN (ablation) | Hier-EgoPack backbone · Ego4D MQ (val) | avg mAP: 19.73 -> 20.21 | Improved |
| add positional encoding to SAGE (ablation) | SAGE · Ego4D MQ (val) | avg mAP: 15.73 -> 17.66 | Improved |
| use 2 TDGC layers per stage instead of 1 (ablation) | Hier-EgoPack · Ego4D MQ (val) | mAP: 18.57 -> 20.21 | Improved |
| use 4 TDGC layers per stage instead of 2 (ablation) | Hier-EgoPack · Ego4D MQ (val) | mAP: 20.21 -> 19.93 | No clear effect |
Closest papers by meaning
- A Backpack Full of Skills: Egocentric Video Understanding with Diverse Task Perspectives (CVPR 2024 · similarity 0.92)
- Learning to Evolve Scenes: Reasoning about Human Activities with Scene Graphs (NeurIPS 2026 · similarity 0.89)
- EgoThinker: Unveiling Egocentric Reasoning with Spatio-Temporal CoT (NeurIPS 2025 · similarity 0.89)
- EgoExoBench: A Benchmark for First- and Third-person View Video Understanding in MLLMs (NeurIPS 2025 · similarity 0.88)
- Omnia de EgoTempo: Benchmarking Temporal Understanding of Multi-Modal LLMs in Egocentric Videos (CVPR 2025 · similarity 0.87)
- Action Scene Graphs for Long-Form Understanding of Egocentric Videos (CVPR 2024 · 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.
