# InterLinK Human-Object Interaction (HOI) Video Dataset The dataset is partially published (the mp4 and RGBD videos are currently available). Due to limitation/errors in uploading the full dataset content in this repository (large files more than 5GB), access to the full content is provided via the following link InterLink HOI dataset The InterLinK HOI dataset is organized by activities (recording sessions) and subjects. Each sequence should includes: - 🎥 Video files in **MP4** formats and SVO RGBD format (by StereoLabs).- 🕺 Mocap skeletons in **BVH**- 🗒️ Text and **JSON** files for annotations and metadata- 🧱 3D CAD or reconstructed/scanned models of objects Information about the InterLinK HOI video dataset will also be hosted in the **GitHub page of the project**: [https://github.com/INTERLINK-HMU](https://github.com/INTERLINK-HMU) --- ## 📁 Dataset Structure The dataset is organized into structured folders per subject, activity, and repetition, ensuring consistent access and traceability. Each sequence directory contains the following: ```videos/├── egocentric/ # Egocentric GoPro video (MP4, 30 fps)├── multiview/ # Exocentric RGB-D video from ZED (MP4, 30 fps RGB + depth aligned)└── RBGD/ # Exocentric RGBD data acquired using a StereoLabs ZED stereo cameramocap/ # 3D human body joint positions (CSV, BVH and JSON formats), including full-body skeletal hierarchiesannotations/ # JSON files for action labels, temporal segments, object and hand masks (binary PNG), and additional metadata3D_objects/ # 3D models of interacted objects in OBJ format with associated texture mapsposes/ # Estimated 3D object poses (in camera and world coordinates), and calibration files (intrinsics, extrinsics)metadata.yaml # Summary of sequence-level attributes (activity type, objects used, repetitions, durations, etc.)``` --- ## 🌐 Availability The dataset is currently hosted on a secure storage unit at the host institution and has also been freely published online via the Zenodo platform/data repository: 🔗 [https://zenodo.org/records/16532005](https://zenodo.org/records/16532005) under a license agreement (**CC BY-NC-SA 4.0**).