---
license: cc-by-nc-4.0
configs:
- config_name: improvised
data_files:
- split: dev
path:
- improvised/dev/**/*
- split: test
path:
- improvised/test/**/*
- split: train
path:
- improvised/train/**/*
- config_name: naturalistic
data_files:
- split: dev
path:
- naturalistic/dev/**/*
- split: test
path:
- naturalistic/test/**/*
- split: train
path:
- naturalistic/train/**/*
tags:
- webdataset
- audio
- video
pretty_name: Seamless Interaction
---
Seamless Interaction Dataset
**A large-scale multimodal dataset of 4,000+ hours of human interactions for AI research**
Human communication involves a complex interplay of verbal and nonverbal signals, essential for conveying meaning and achieving interpersonal goals.
The **Seamless Interaction Dataset** is a large-scale collection of over 4,000 hours of face-to-face interaction footage from more than 4,000 participants in diverse contexts.
This dataset enables the development of AI technologies that understand human interactions and communication, unlocking breakthroughs in:
- ๐ค Virtual agents and embodied AI
- ๐ญ Natural human-computer interaction
- ๐ก Advanced telepresence experiences
- ๐ Multimodal content analysis tools
- ๐ฌ Animation and synthetic content generation
## ๐ Quick Start
```bash
git clone https://github.com/facebookresearch/seamless-interaction
cd seamless-interaction
pip install -e .
streamlit run src/seamless_interaction/app/Welcome.py
# if you use uv
uv sync
uv run streamlit run src/seamless_interaction/app/Welcome.py
```
Explore the dataset with our interactive browser:
**Features:**
- ๐ **Hierarchical Navigation**: Browse by Label โ Split โ Batch โ Interaction
- ๐ฒ **Random Sampling**: Discover interactions with one-click random selection
- ๐ฅ **Download Interface**: Download specific batches with size estimation and progress tracking
- ๐ฌ **Video Viewer**: Side-by-side participant videos with synchronized playback
- ๐ **Data Analysis**: Overview statistics and distribution plots
- ๐ **File Management**: Organize and preview audio, JSON, and NPZ files with expandable dropdowns
### Download Options
We provide comprehensive download methods supporting all research scales and requirements:
| **Scale** | **Size** | **Method** | **Use Case** | **Script** | **Sampling** |
|-----------|----------|------------|--------------|------------|-------------|
| ๐ **Single Example** | ~100MB | S3 | Quick exploration, understanding data structure | [`download_s3.py`](https://github.com/facebookresearch/seamless_interaction/blob/main/scripts/download_s3.py#L10) | Auto-sample from preferred vendors |
| ๐ฅ **Interaction Pair** | ~200MB | S3 | Study conversational dynamics between participants | [`download_s3.py`](https://github.com/facebookresearch/seamless_interaction/blob/main/scripts/download_s3.py#L34) | Auto-detect conversation pairs |
| ๐ **Sample Set** | ~1GB | S3/HF | Initial prototyping, algorithm development | [`download_s3.py`](https://github.com/facebookresearch/seamless_interaction/blob/main/scripts/download_s3.py#L66), [`download_hf.py`](https://github.com/facebookresearch/seamless_interaction/blob/main/scripts/download_hf.py#L10) | File selection or archive-based |
| ๐ฏ **Session Groups** | ~400MB | S3 | Deep conversational context, session dynamics | [`download_s3.py`](https://github.com/facebookresearch/seamless_interaction/blob/main/scripts/download_s3.py#L100) | Auto-sample rich sessions |
| ๐ฆ **Single Batch** | ~50GB | HF | Substantial local development, full exploration | [`download_hf.py`](https://github.com/facebookresearch/seamless_interaction/blob/main/scripts/download_hf.py#L24) | WebDataset tarball download |
| ๐๏ธ **Multiple Batches** | ~150GB+ | HF | Training datasets, large-scale analysis | [`download_hf.py`](https://github.com/facebookresearch/seamless_interaction/blob/main/scripts/download_hf.py#L38) | WebDataset tarball download |
| ๐ฏ **Different Splits** | Variable | HF | Cross-validation (train/dev/test, improvised/naturalistic) | [`download_hf.py`](https://github.com/facebookresearch/seamless_interaction/blob/main/scripts/download_hf.py#L55) | WebDataset tarball download |
| ๐ **Whole Dataset** | ~27TB | HF | Complete research dataset, production systems | [`download_hf.py`](https://github.com/facebookresearch/seamless_interaction/blob/main/scripts/download_hf.py#L82) | WebDataset tarball download |
### Basic Data Loading (HF + WebDataset)
```python
from datasets import load_dataset
# configure
label = "improvised"
split = "dev"
batch_idx = 0
archive_list = [0, 1]
base_url = (
f"https://huggingface.co/datasets/facebook/"
f"seamless-interaction/resolve/main/{label}/{split}/"
"{batch_idx:04d}/{archive_idx:04d}.tar"
)
urls = [base_url.format(batch_idx=batch_idx, archive_idx=archive_idx) for archive_idx in archive_list]
dataset = load_dataset(
"webdataset", data_files={split: urls}, split=split, streaming=True
)
for item in dataset:
break
isinstance(item["mp4"], bytes)
# True
item["npz"].keys()
# dict_keys(['boxes_and_keypoints:box', 'boxes_and_keypoints:is_valid_box', 'boxes_and_keypoints:keypoints', 'movement:EmotionArousalToken', 'movement:EmotionValenceToken', 'movement:FAUToken', 'movement:FAUValue', 'movement:alignment_head_rotation', 'movement:alignment_translation', 'movement:emotion_arousal', 'movement:emotion_scores', 'movement:emotion_valence', 'movement:expression', 'movement:frame_latent', 'movement:gaze_encodings', 'movement:head_encodings', 'movement:hypernet_features', 'movement:is_valid', 'smplh:body_pose', 'smplh:global_orient', 'smplh:is_valid', 'smplh:left_hand_pose', 'smplh:right_hand_pose', 'smplh:translation'])
item["json"].keys()
# dict_keys(['id', 'metadata:transcript', 'metadata:vad'])
item["wav"].keys()
# dict_keys(['path', 'array', 'sampling_rate'])
```
## ๐ฆ Deep Dive into the Dataset
### Dataset Structure
The Seamless Interaction Dataset is organized into two main categories/labels:
- **Improvised**: Interactions primarily based on predefined scenarios with guided prompts with at least one professional actor.
- **Naturalistic**: Prompted conversations that can be carried out by normal people.
```
seamless_interaction
โโโ interactions.csv # Metadata for prompts
โโโ participants.csv # Metadata for participants
โโโ relationships.csv # Metadata for participant relationships per session
โโโ improvised # Interactions with guided prompts
โ โโโ dev
โ โ โโโ 1P-IS/ # First-party internal state annotations
โ โ โ โโโ V_S_I_P.json
โ โ โโโ 1P-R/ # First-party internal state rationale annotations
โ โ โ โโโ V_S_I_P.json
โ โ โโโ 3P-IS/ # Third-party internal state annotations
โ โ โ โโโ V_S_I_P.json
โ โ โโโ 3P-R/ # Third-party internal state rationale annotations
โ โ โ โโโ V_S_I_P.json
โ โ โโโ 3P-V/ # Third-party visual annotation
โ โ โ โโโ V_S_I_P.json
โ โ โโโ audio/ # Speaker-bleed denoised audio
โ โ โ โโโ V_S_I_P.wav
โ โ โโโ boxes_and_keypoints/
โ โ โ โโโ box/ # Bounding boxes for each participant
โ โ โ โโโ is_valid_box/ # Whether bounding boxes are valid
โ โ โ โโโ keypoints/ # Detected facial/body keypoints
โ โ โโโ movement/ # Quantified Imitator movement features
โ โ โ โโโ emotion_arousal/ # Arousal measures
โ โ โ โโโ emotion_valence/ # Valence measures
โ โ โ โโโ emotion_scores/ # Emotion detection scores
โ โ โ โโโ expression/ # Facial expression parameters
โ โ โ โโโ FAUToken/ # Facial Action Unit tokens
โ โ โ โโโ FAUValue/ # Facial Action Unit values
โ โ โ โโโ gaze_encodings/ # Eye gaze direction encodings
โ โ โ โโโ head_encodings/ # Head position/rotation encodings
โ โ โ โโโ frame_latent/ # Per-frame latent representations
โ โ โ โโโ is_valid/ # Validity flags for extracted features
โ โ โโโ smplh/ # SMPL-H body model parameters
โ โ โ โโโ body-pose/ # Body pose parameters
โ โ โ โโโ global_orient/ # Global orientation parameters
โ โ โ โโโ is_valid/ # Valid frames indicators
โ โ โ โโโ left_hand_pose/ # Left hand pose parameters
โ โ โ โโโ right_hand_pose/ # Right hand pose parameters
โ โ โ โโโ translation/ # Global translation parameters
โ โ โโโ transcript/ # Time-aligned speech transcription
โ โ โ โโโ V_S_I_P.jsonl
โ โ โโโ vad/ # Voice activity detection
โ โ โ โโโ V_S_I_P.jsonl
โ โ โโโ video/ # Raw HD video recordings
โ โ โโโ V_S_I_P.mp4
โ โโโ test/ # Test split with similar structure
โ โโโ train/ # Training split with similar structure
โโโ naturalistic/ # Spontaneous conversations
โโโ dev/ # Same structure as improvised/dev
โโโ test/ # Same structure as improvised/test
โโโ train/ # Same structure as improvised/train
```
Each file is named according to a consistent convention:
- `V`: Collection site/vendor identifier
- `S`: Unique session identifier
- `I`: Specific interaction within a session
- `P`: Individual participant identifier
### Available Modalities and Features
Each interaction in the dataset includes:
| Modality | Description | File Format | Sample Rate |
|----------|-------------|-------------|-------------|
| ๐ฅ Video | High-definition face-to-face footage | MP4 (H.264) | 30/29.97 FPS, 1080p |
| ๐๏ธ Audio | Denoised audio with separate channels | WAV | 48kHz, 16-bit |
| ๐ Transcript | Time-aligned speech transcription | JSONL | - |
| ๐ SMPL-H | 3D body model parameters | NPY | 30 Hz |
| ๐ง Imitator Movement Features | Comprehensive quantified imitator movement data | NPY | 30 Hz |
| ๐ Annotations | Human-annotated behavioral data | JSON | - |
| ๐ VAD | Voice activity detection | JSONL | 100 Hz |
| ๐ฆ Keypoints | Face and body keypoints | NPY | 30 Hz |
#### Annotation Types
The dataset includes several types of human annotations for rich behavioral analysis:
| Annotation | Hours | Total Annotations | Mean # Tokens |
|------------|-------------|--------|--------|
| 1P-IS (1st-party internal state annotations) | 1.1 | 751 | 5.8 |
| 1P-R (1st-party internal state rationale annotations) | 1.1 | 751 | 10.2 |
| 3P-IS (3rd-party internal state annotations) | 4.7 | 5132 | 5.2 |
| 3P-R (3rd-party internal state rationale annotations) | 4.7 | 5132 | 11.3 |
| 3P-V (3rd-party visual annotation) | 4.7 | 5132 | 14.6 |
Please refer to the [technical report](https://ai.meta.com/research/publications/seamless-interaction-dyadic-audiovisual-motion-modeling-and-large-scale-dataset/) for a more detailed overview of annotations.
#### Movement/Imitator Feature Types
The movement directory contains rich behavioral features (output of the Imitator model):
| Feature | Description |
|---------|-------------|
| `emotion_arousal` | Arousal intensity measurements |
| `emotion_valence` | Valence (positive/negative) measurements |
| `emotion_scores` | Detected emotion categorical scores |
| `expression` | Parametric facial expression encodings |
| `FAUToken`/`FAUValue` | Facial Action Unit tokens and intensity values |
| `gaze_encodings` | Neural encodings of gaze direction |
| `head_encodings` | Neural encodings of head position and rotation |
| `frame_latent` | Per-frame latent representations |
| `alignment_head_rotation` | Head rotation data for temporal alignment |
| `alignment_translation` | Translation parameters for temporal alignment |
| `EmotionArousalToken`/`EmotionValenceToken` | Discretized emotion tokens |
| `hypernet_features` | Features from hypernetwork processing |
### Dataset Versions
The dataset is organized in self-contained batches for flexible exploration:
| Split | Batches | Size per Batch | Total Size | Description |
|-------|---------|----------------|------------|-------------|
| **dev** | 5 | ~50GB | ~500GB | Development/validation set |
| **test** | 5 | ~50GB | ~500TB | Hold-out test set |
| **train** | 200+ | ~50GB | ~20TB+ | Full training data |
#### File Format Specifications
Our data is stored in the following formats for optimal usability:
| Format | Description | Usage |
|--------|-------------|-------|
| NPZ | NumPy array files | Efficient storage of numerical feature vectors, keypoints, and parameters |
| JSONL | JSON Lines | Time-aligned annotations with one event per line (e.g., transcripts, VAD) |
| JSON | JavaScript Object Notation | Structured metadata and annotations with timestamps |
| MP4 | MPEG-4 Part 14 | High-quality compressed video with H.264 encoding |
| WAV | Waveform Audio | Uncompressed audio for highest fidelity processing |
## ๐งช Research Applications
The Seamless Interaction Dataset enables research across multiple domains:
### Embodied AI and Virtual Agents
- Train agents that display natural gestures
- Model turn-taking dynamics and interaction rhythms
- Generate contextually appropriate responses to human behavior
### Multimodal Understanding
- Analyze cross-modal correlations between speech, gesture, and expressions
- Extract behavioral patterns from large-scale interaction data
- Develop models to understand social dynamics
### Human-Computer Interaction
- Design interfaces that respond to subtle human cues
- Improve telepresence technologies with better behavioral modeling
- Create more natural conversational agents
### Animation and Content Creation
- Generate realistic human behaviors for animated characters
- Synthesize conversational dynamics for virtual production
- Create training data for digital human technologies
## โ ๏ธ Known Limitations and Noise in Metadata
Given the scale and complexity involved in collecting the Seamless Interaction dataset, there are several known limitations that we will address in our ongoing work, with improvements planned for in future versions:
### Errors in Human-Based Time-Stamping
The core unit of the dataset is interactions. An interaction defines the *active time* during which a participantโs conversation and behavior can be linked to a pair of prompts. We have observed instances of misaligned time-stamps, including:
- Annotated start/end times may be too early or too late.
- Occasional misalignment between prompt text and spoken material.
- Ordering of prompts that may contain off-by-one errors.
Despite our efforts to automatically identify and correct these errors, approximately 10% of the interactions remain affected.
### Time Stamping "Noise" in Moments of Interest (MOI)
While defining a MOI inherently involves some subjectivity, there are rare instances where:
- The described behavior only represents a subset of the observed behavior.
- The duration of the MOI does not fully capture the annotated behavior.
### Incorrect Assignment of Participant IDs
In rare instances, we have observed:
- Duplicate participant identifiers being assigned to different individuals.
- The same individual being mapped to different identifiers.
### Unreleased "Meta Time"
Currently, the dataset only contains *active time* segments - time in which two participants are actively responding to prompts. *Meta time* refers to the time between *active segments* in which participants are studying their new prompts, taking a break, etc. *Meta time* constitutes hundreds of hours in the raw collection and maybe be explored for future releases.
### Variation in Recording Site Consistency
This multi-site project contains variation in:
- Recording quality, including issues like speaker bleed and participants staying in frame.
- Acting quality in *Improvised* segments.
- The likelihood of time-stamping errors.
All vendors met our technical requirements; however,there is noticeable variation in production quality across different sites.
## ๐ License & Data Usage Policy
The Seamless Interaction Dataset is licensed under CC-BY-NC 4.0 (Creative Commons Attribution-NonCommercial 4.0 International).
This means you are free to:
- **Share** โ copy and redistribute the material in any medium or format
- **Adapt** โ remix, transform, and build upon the material
Under the following terms:
- **Attribution** โ You must give appropriate credit, provide a link to the license, and indicate if changes were made.
- **NonCommercial** โ You may not use the material for commercial purposes without explicit permission.
## ๐ Citation
If you use the Seamless Interaction Dataset in your research, please cite:
BibTeX
```bibtex
@article{seamless_interaction,
title={Seamless Interaction: Dyadic Audiovisual Motion Modeling and Large-Scale Dataset},
author={Vasu Agrawal and
Akinniyi Akinyemi and
Kathryn Alvero and
Morteza Behrooz and
Julia Buffalini and
Fabio Maria Carlucci and
Joy Chen and
Junming Chen and
Zhang Chen and
Shiyang Cheng and
Praveen Chowdary and
Joe Chuang and
Antony D'Avirro and
Jon Daly and
Ning Dong and
Mark Duppenthaler and
Cynthia Gao and
Jeff Girard and
Martin Gleize and
Sahir Gomez and
Hongyu Gong and
Srivathsan Govindarajan and
Brandon Han and
Sen He and
Denise Hernandez and
Yordan Hristov and
Rongjie Huang and
Hirofumi Inaguma and
Somya Jain and
Raj Janardhan and
Qingyao Jia and
Christopher Klaiber and
Dejan Kovachev and
Moneish Kumar and
Hang Li and
Yilei Li and
Pavel Litvin and
Wei Liu and
Guangyao Ma and
Jing Ma and
Martin Ma and
Xutai Ma and
Lucas Mantovani and
Sagar Miglani and
Sreyas Mohan and
Louis-Philippe Morency and
Evonne Ng and
Kam-Woh Ng and
Tu Anh Nguyen and
Amia Oberai and
Benjamin Peloquin and
Juan Pino and
Jovan Popovic and
Omid Poursaeed and
Fabian Prada and
Alice Rakotoarison and
Alexander Richard and
Christophe Ropers and
Safiyyah Saleem and
Vasu Sharma and
Alex Shcherbyna and
Jia Shen and
Jie Shen and
Anastasis Stathopoulos and
Anna Sun and
Paden Tomasello and
Tuan Tran and
Arina Turkatenko and
Bo Wan and
Chao Wang and
Jeff Wang and
Mary Williamson and
Carleigh Wood and
Tao Xiang and
Yilin Yang and
Zhiyuan Yao and
Chen Zhang and
Jiemin Zhang and
Xinyue Zhang and
Jason Zheng and
Pavlo Zhyzheria and
Jan Zikes and
Michael Zollhoefer
},
url={https://ai.meta.com/research/publications/seamless-interaction-dyadic-audiovisual-motion-modeling-and-large-scale-dataset/},
year={2025}
}
```
## ๐ Acknowledgments
This project was made possible thanks to contributions from:
- The thousands of participants who provided interaction data
- Our dedicated annotation and QA team
- Research collaborators from multiple institutions
- FAIR (Fundamental AI Research)
- The open-source community for valuable tools and libraries
- Our data collection partners across multiple sites
- Meta Reality Labs for supporting this research initiative