Ego-4D

Ego4D is a large-scale, wearable-egocentric dataset that contains video recordings of everyday activities, used to evaluate proactive procedural assistance systems and their ability to handle deviations from expected task sequences.

Papers using Ego-4D (9)

Plan, Watch, Recover: A Benchmark and Architectures for Proactive Procedural Assistance2026Recurrent Reasoning with Vision-Language Models for Estimating Long-Horizon Embodied Task Progress2026EgoSound: Benchmarking Sound Understanding in Egocentric Videos2026EVA02-AT: Egocentric Video-Language Understanding with Spatial-Temporal Rotary Positional Embeddings and Symmetric Optimization2025Know-show: Benchmarking Video-language Models On Spatio-temporal Grounded Reasoning2025Vision and Intention Boost Large Language Model in Long-Term Action Anticipation2025HourVideo: 1-Hour Video-Language Understanding2024 · 1 citesX-MIC: Cross-Modal Instance Conditioning for Egocentric Action Generalization2024VLM Agents Generate Their Own Memories: Distilling Experience into Embodied Programs of Thought2024