Overview

This dataset contains 1206 images of people wearing Personal Protective Equipment (PPE) in various environments, such as construction sites, industrial workplaces, and more. The goal of this dataset is to provide valuable resources to train Computer Vision models capable of identifying safety equipment.

Model Training

If you'd like to train an Object Detection Model with this dataset, you can follow the step by step guide in our tensorflow-object-detection GitHub repository

Classes and Annotation Format

The dataset includes 7.724 annotations across the following object classes:

• helmet - 2.543 objects;
• no-helmet - 129 objects;
• no-vest - 892 objects;
• person - 2.817 objects;
• vest - 1.343 objects;

The dataset is annotated in the YOLO annotation format, in which each annotation corresponds to a single line in a text file, with each line representing one object instance. The format for each line is as follows:

<object-class> <x_center> <y_center> <width> <height>

• <object-class>: The integer index representing the object's class based on the order they appear in the classes.txt file.

• <x_center>, <y_center>: The normalized coordinates of the center of the bounding box with respect to the width and height of the image. Both <x_center> and <y_center> should be in the range [0, 1]. The top-left corner of the image is (0,0), and the bottom-right corner is (1,1).

• <width>, <height>: The normalized width and height of the bounding box, also relative to the width and height of the image. Both <width> and <height> should be in the range [0, 1].

For any questions or issues related to the dataset, please contact contact@amlstation.com.

🔗 Visit our website: www.amlstation.com

🔗 Follow us on LinkedIn: www.linkedin.com/company/amlstation

License

License: CC BY 4.0

This dataset is based on Worker-Safety Computer Vision Project, which is licensed under a Creative Commons Attribution 4.0 International License.

The following changes were made to the original dataset:

• All images and annotations were placed inside the images folder without being divided into train, test and validation sets;
• Classes names, originally placed inside _darknet.labels files, were placed in classes.txt;

You can find the complete license text at the Creative Commons website: https://creativecommons.org/licenses/by/4.0/