Egocentric-100K

Egocentric-100K is the largest dataset of manual labor. You can visualize the dataset here. Egocentric-100K is state-of-the-art in hand visibility and active manipulation density compared to previous in-the-wild egocentric datasets. The complete 30,000 frame evaluation set is available at Egocentric-100K-Evaluation. Dataset Statistics Attribute Value Total Hours 100,405 Total Frames 10.8 billion Video Clips 2,010,759 Median Clip Length 180.0 seconds Mean Hours per Worker 7.06 Storage Size 24.79 TB Format H.265/MP4 Resolution 256p (456x256) Frame Rate 30 fps Camera Type Monocular head-mounted fisheye Audio No Device Build AI Gen 1 Camera Intrinsics Each worker folder contains an intrinsics.json file with calibrated camera parameters for that worker's device. The intrinsics use the OpenCV fisheye model (Kannala-Brandt equidistant projection) with 4 distortion coefficients (k1-k4). All values are scaled appropriately for the 456x256 resolution. Example intrinsics.json: { "model": "fisheye", "image_width": 456, "image_height": 256, "fx": 137.98, "fy": 138.23, "cx": 232.17, "cy": 125.37, "k1": 0.3948, "k2": 0.1798, "k3": -0.2753, "k4": 0.0793 } Dataset Structure Egocentric-100K is structured in WebDataset format: builddotai/Egocentric-100K/ ├── factory001/ │ ├── worker001/ │ │ ├── intrinsics.json # Camera intrinsics for this worker │ │ ├── part000.tar # Shard 0 (≤1GB) │ │ └── part001.tar # Shard 1 (if needed) │ ├── worker002/ │ │ ├── intrinsics.json │ │ └── part000.tar │ └── ... │ ├── factory002/ │ ├── worker001/ │ │ ├── intrinsics.json │ │ └── part000.tar │ └── ... │ └── ... Each TAR file contains pairs of video and metadata files: part000.tar ├── factory001_worker001_00001.mp4 # Video 1 ├── factory001_worker001_00001.json # Metadata for video 1 ├── factory001_worker001_00002.mp4 # Video 2 ├── factory001_worker001_00002.json # Metadata for video 2 └── ... # Additional video/metadata pairs Each JSON metadata file has the following fields: { "factory_id": "factory_002", "