| --- |
| language: |
| - en |
| license: cc-by-4.0 |
| tags: |
| - multiple-modality |
| - industrial-scene |
| - custom-dataset |
| size_categories: |
| - 10G<n<100G |
| source_datasets: |
| - original |
| --- |
| |
| # InspecSafe-V1 |
|
|
| ## Overview |
|
|
| **InspecSafe-V1** is a high-quality, multimodal annotated dataset designed for **world model construction and analysis in industrial environments**. The data was collected from real-world inspection robots deployed across industrial sites and has been carefully cleaned and standardized for research and applications in predictive world modeling for industrial scenarios. |
|
|
| The dataset covers five representative industrial settings: tunnels, power facilities, sintering equipment, oil/gas/chemical plants, and coal conveyor galleries. It was constructed using data from 41 wheeled or rail-mounted inspection robots operating at 2,239 valid inspection waypoints. Across the dataset, multimodal records may include visible-light video, infrared video, audio, depth or LiDAR point clouds, gas concentration readings, temperature, and humidity. Depending on the inspection robot, sensing configuration, and waypoint conditions, each inspection waypoint is associated with the available subset of these modalities rather than necessarily containing all modality types. |
| The available modality types include: |
|
|
| - Visible-light video |
| - Infrared video |
| - Audio |
| - Depth or LiDAR point clouds |
| - Gas concentration readings |
| - Temperature |
| - Humidity |
|
|
| |
| Additionally, pixel-level polygonal segmentation annotations are provided for industrial objects in visible-light images. To support downstream tasks, each sample is also accompanied by a semantic scene description and a corresponding safety-level label based on real inspection protocols. |
|
|
| ## Dataset Format |
|
|
| The dataset is divided into a training set and a test set, both of which are organized in a structured directory layout with aligned multimodal streams and annotations. An overview of the data structure is shown below: |
|
|
|
|
| ``` |
| DATA_PATH |
| ├── train |
| │ ├── Annotations |
| │ │ ├── Normal_data |
| │ │ │ ├── coal_conveyor-Level04-SuspendedRail-000560 |
| │ │ │ │ ├── coal_conveyor-Level04-SuspendedRail-000560-001.jpg |
| │ │ │ │ ├── coal_conveyor-Level04-SuspendedRail-000560-001.json |
| │ │ │ │ └── coal_conveyor-Level04-SuspendedRail-000560-001.txt |
| │ │ │ └── ... |
| │ │ └── Anomaly_data |
| │ │ ├── coal_conveyor-Level01-SuspendedRail-002486 |
| │ │ │ ├── coal_conveyor-Level01-SuspendedRail-002486-001.jpg |
| │ │ │ ├── coal_conveyor-Level01-SuspendedRail-002486-001.json |
| │ │ │ └── coal_conveyor-Level01-SuspendedRail-002486-001.txt |
| │ │ └── ... |
| │ ├── Other_modalities |
| │ │ ├── coal_conveyor-Level04-SuspendedRail-000560 |
| │ │ │ ├── coal_conveyor-Level04-SuspendedRail-000560-visible.mp4 |
| │ │ │ ├── coal_conveyor-Level04-SuspendedRail-000560-infrared.mp4 |
| │ │ │ ├── coal_conveyor-Level04-SuspendedRail-000560-sensor.txt |
| │ │ │ ├── coal_conveyor-Level04-SuspendedRail-000560-point_cloud.bag |
| │ │ │ └── coal_conveyor-Level04-SuspendedRail-000560-audio.wav |
| │ │ └── ... |
| │ └── Parameters |
| │ ├── Hardware |
| │ ├── Device_A.json |
| │ ├── Device_B.json |
| │ └── ... |
| └── test |
| ├── Annotations |
| │ ├── Normal_data |
| │ │ ├── coal_conveyor-Level04-SuspendedRail-000001 |
| │ │ │ ├── coal_conveyor-Level04-SuspendedRail-000001-001.jpg |
| │ │ │ ├── coal_conveyor-Level04-SuspendedRail-000001-001.json |
| │ │ │ └── coal_conveyor-Level04-SuspendedRail-000001-001.txt |
| │ │ └── ... |
| │ └── Anomaly_data |
| │ ├── coal_conveyor-Level01-SuspendedRail-002235 |
| │ │ ├── coal_conveyor-Level01-SuspendedRail-002235-001.jpg |
| │ │ ├── coal_conveyor-Level01-SuspendedRail-002235-001.json |
| │ │ └── coal_conveyor-Level01-SuspendedRail-002235-001.txt |
| │ └── ... |
| ├── Other_modalities |
| │ ├── coal_conveyor-Level04-SuspendedRail-000001 |
| │ │ ├── coal_conveyor-Level04-SuspendedRail-000001-visible.mp4 |
| │ │ ├── coal_conveyor-Level04-SuspendedRail-000001-infrared.mp4 |
| │ │ ├── coal_conveyor-Level04-SuspendedRail-000001-sensor.txt |
| │ │ ├── coal_conveyor-Level04-SuspendedRail-000001-point_cloud.bag |
| │ │ └── coal_conveyor-Level04-SuspendedRail-000001-audio.wav |
| │ └── ... |
| └── Parameters |
| ├── Hardware |
| ├── Device_A.json |
| ├── Device_B.json |
| └── ... |
| ``` |
|
|
|
|
| ### Notes: |
| - **Inspection point identifier**: Each folder name represents an inspection point, such as `coal_conveyor-Level04-SuspendedRail-000560`. The same identifier is used in both the `Annotations` and `Other_modalities` folders to enable cross-modal correspondence. |
| - **Inspection instances**: A single inspection point may contain multiple inspection instances. In the `Annotations` folder, these instances are distinguished by numerical suffixes appended to the filenames, such as `-001`, `-002`, and `-003`. |
| - **Annotations**: |
| - `.jpg`: Visible-light image frame for the corresponding inspection instance. |
| - `.json`: Pixel-level polygonal segmentation annotations and related metadata. |
| - `.txt`: Human-readable semantic description of the scene. |
| - **Other modalities**: |
| - `.mp4`: Visible-light and infrared videos. |
| - `.txt`: Sensor logs, including gas concentration, temperature, and humidity. |
| - `.bag`: Point-cloud data in ROS bag format. |
| - `.wav`: Audio recordings. |
| - **Parameters**: The `Parameters` folder contains hardware specifications, software settings, and calibration-related files used to support multimodal interpretation and fusion. |
| > This structure ensures synchronized access across all modalities and supports both supervised learning and world modeling tasks. Each sample metadata (e.g., robot ID, location, timestamp, safety label) is stored in JSON format. Segmentation masks are provided as PNG images with instance IDs matching the annotation JSON. |
|
|
| The shared inspection point identifier allows the multimodal and sensory records in `Other_modalities` to be linked to one or more annotated inspection instances in `Annotations`. |
|
|
| ## Data Splits |
|
|
| | Split | Number of Samples | |
| |---------|-------------------| |
| | train | 3,763 | |
| | test | 1,250 | |
|
|
| > Note: The dataset does not include a separate validation split; users are encouraged to create one from the training set as needed. |
|
|
| ## License |
|
|
| This dataset is released under the [CC-BY-4.0 License](https://creativecommons.org/licenses/by/4.0/). |
|
|
| ## Acknowledgements |
|
|
| We thank the multimodal recognition algorithm team who contributed to data collection and annotation. This work was supported by TetraBOT. |
|
|
| --- |
|
|
| For questions or contributions, please open an issue in the repository. |
