HumanNet: Scaling Human-centric Video Learning to One Million Hours

2026-05-07

AI Summary

The scarcity of large-scale, diverse human activity video datasets with rich interaction annotations has significantly hindered the advancement of embodied intelligence in physical interaction learning. To address this, this work proposes a systematic data construction paradigm for embodied learning and introduces HumanNet, a million-hour-scale egocentric video corpus encompassing both first- and third-person perspectives. HumanNet covers fine-grained activities, human-object interactions, tool usage, and long-horizon behaviors, accompanied by interaction-aware multimodal annotations. Leveraging this dataset, a Qwen vision-language model fine-tuned on only 1,000 hours of first-person video outperforms Magic Cobot—a model trained on 100 hours of real robot data—on a fixed validation benchmark, demonstrating that human demonstration videos serve as an effective, scalable alternative for robot learning.

Abstract

Progress in embodied intelligence increasingly depends on scalable data infrastructure. While vision and language have scaled with internet corpora, learning physical interaction remains constrained by the lack of large, diverse, and richly annotated human activity data. We present HumanNet, a one-million-hour human-centric video corpus that captures how humans interact with the physical world at scale. HumanNet spans both first-person and third-person perspectives and covers fine-grained activities, human-object interactions, tool use, and long-horizon behaviors across diverse real-world environments. Beyond raw video, the dataset provides interaction-centric annotations, including captions, motion descriptions, and hand and body-related signals, enabling motion-aware and interaction-aware learning. Beyond scale, HumanNet introduces a systematic data curation paradigm for embodied learning, where human-centric filtering, temporal structuring, viewpoint diversity, and annotation enrichment are treated as first-class design principles. This design transforms unstructured internet video into a scalable substrate for representation learning, activity understanding, motion generation, and human-to-robot transfer. We conduct a first-step validation on the value of this design through controlled vision-language-action ablation: under a fixed set of validation data, continued training from the Qwen VLM model with 1000 hours of egocentric video drawn from HumanNet surpasses the continued training with 100 hours of real-robot data from Magic Cobot, indicating that egocentric human video could be a scalable and cost-effective substitute for robot data. By building this project, we aim to explore the opportunity to scale embodied foundation models using human-centric videos, rather than relying solely on robot-specific data.

Problem

• embodied intelligence
• human-centric video
• human-object interaction
• scalable data
• activity understanding

Innovation

• Human-centric video
• embodied learning
• scalable dataset
• interaction-aware annotation
• egocentric vision