Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

SH17: A Dataset for Human Safety and Personal Protective Equipment Detection in Manufacturing Industry

Hafiz Mughees Ahmad, Afshin Rahimi

TL;DR

This work addresses PPE compliance in manufacturing by introducing SH17, a diverse, open-source dataset with 8,099 images and 75,994 instances across 17 classes to train and benchmark object detectors. It benchmarks multiple YOLO-based detectors (notably Yolov9-e) on SH17, achieving 70.9% mAP@50 and demonstrating competitive performance with fewer trainable parameters. The authors also show cross-domain generalization by testing Yolov9-e on the Pictor-PPE dataset, validating practical applicability for industry. Overall, SH17 provides a comprehensive benchmark and a path toward scalable, non-invasive PPE compliance in manufacturing environments.

Abstract

Workplace accidents continue to pose significant risks for human safety, particularly in industries such as construction and manufacturing, and the necessity for effective Personal Protective Equipment (PPE) compliance has become increasingly paramount. Our research focuses on the development of non-invasive techniques based on the Object Detection (OD) and Convolutional Neural Network (CNN) to detect and verify the proper use of various types of PPE such as helmets, safety glasses, masks, and protective clothing. This study proposes the SH17 Dataset, consisting of 8,099 annotated images containing 75,994 instances of 17 classes collected from diverse industrial environments, to train and validate the OD models. We have trained state-of-the-art OD models for benchmarking, and initial results demonstrate promising accuracy levels with You Only Look Once (YOLO)v9-e model variant exceeding 70.9% in PPE detection. The performance of the model validation on cross-domain datasets suggests that integrating these technologies can significantly improve safety management systems, providing a scalable and efficient solution for industries striving to meet human safety regulations and protect their workforce. The dataset is available at https://github.com/ahmadmughees/sh17dataset.

SH17: A Dataset for Human Safety and Personal Protective Equipment Detection in Manufacturing Industry

TL;DR

This work addresses PPE compliance in manufacturing by introducing SH17, a diverse, open-source dataset with 8,099 images and 75,994 instances across 17 classes to train and benchmark object detectors. It benchmarks multiple YOLO-based detectors (notably Yolov9-e) on SH17, achieving 70.9% mAP@50 and demonstrating competitive performance with fewer trainable parameters. The authors also show cross-domain generalization by testing Yolov9-e on the Pictor-PPE dataset, validating practical applicability for industry. Overall, SH17 provides a comprehensive benchmark and a path toward scalable, non-invasive PPE compliance in manufacturing environments.

Abstract

Workplace accidents continue to pose significant risks for human safety, particularly in industries such as construction and manufacturing, and the necessity for effective Personal Protective Equipment (PPE) compliance has become increasingly paramount. Our research focuses on the development of non-invasive techniques based on the Object Detection (OD) and Convolutional Neural Network (CNN) to detect and verify the proper use of various types of PPE such as helmets, safety glasses, masks, and protective clothing. This study proposes the SH17 Dataset, consisting of 8,099 annotated images containing 75,994 instances of 17 classes collected from diverse industrial environments, to train and validate the OD models. We have trained state-of-the-art OD models for benchmarking, and initial results demonstrate promising accuracy levels with You Only Look Once (YOLO)v9-e model variant exceeding 70.9% in PPE detection. The performance of the model validation on cross-domain datasets suggests that integrating these technologies can significantly improve safety management systems, providing a scalable and efficient solution for industries striving to meet human safety regulations and protect their workforce. The dataset is available at https://github.com/ahmadmughees/sh17dataset.
Paper Structure (24 sections, 3 equations, 6 figures, 8 tables)

This paper contains 24 sections, 3 equations, 6 figures, 8 tables.

Table of Contents

  1. Introduction
  2. Literature Review
  3. Existing Datasets
  4. Hardhat Wearing Detection (GDUT-HWD) Dataset
  5. Safety Helmet Wearing (SHW) Dataset
  6. Color Helmet and Vest (CHV) Dataset
  7. SHEL5K Dataset
  8. Pictor-PPE Dataset
  9. TCRSF Dataset
  10. YOLO Models
  11. od for ppe Detection
  12. SH17 Dataset
  13. Collection Process
  14. Annotation Process
  15. Categories
  16. ...and 9 more sections

Figures (6)

  • Figure 1: Samples from the proposed SH17 Dataset. Best viewed online.
  • Figure 2: Distribution of all class instances. The percentage of each class is shown at the top of each bar.
  • Figure 3: Training metrics of yolov9-e model that performs best among all.
  • Figure 4: Confusion Matrix of all the instances of Pictor-PEE dataset.
  • Figure 5: Visual representation of the predictions made by different models; yolov8-n, yolov8-m, yolov8-x, yolov9-e. Best viewed online.
  • ...and 1 more figures