Muhammad Hadif, Dzulkhissham (2022) Real-Time Detection of Personal Protective Equipment for Site Safety Using Deep Learning Techniques. College of Engineering, Universiti Malaysia Pahang Al-Sultan Abdullah.
|
Pdf
EA18167_Hadif_Thesis - Hadif.pdf - Accepted Version Download (2MB) | Preview |
Abstract
Traumatic brain injuries (from falls and electrocution), sprains, broken bones, and other injuries can result from slipping and falling on the ground, leaking gas that is hazardous to inhale and collisions are the primary causes of construction fatalities (resulting from being struck by objects). The Department of Occupational Safety and Health (DOSH) in Malaysia mandates contractors to always enforce and monitor adequate Personal Protective Equipment (PPE) for workers (e.g., hard helmet and vest) as a preventative measure. In addition, because of the COVID-19 outbreak over the last two years, wearing a face mask in factories, departments, and working offices is critical. This paper presents a deep learning technique for detecting multiple personal protection equipment at once based on You-Only-Look-Once Version 4 (YOLOv4) object detection algorithm. The whole training process or computation is done in Google Colaboratory. The training result shows that the Mean Average Precision (mAP) for the best weight training is up to 97.04% for detecting multiple PPE by using this method.
| Item Type: | Undergraduates Project Papers |
|---|---|
| Additional Information: | SV: Ikhwan Hafiz bin Muhamad |
| Uncontrolled Keywords: | construction fatalities, You-Only-Look-Once Version 4 (YOLOv4) |
| Subjects: | T Technology > TA Engineering (General). Civil engineering (General) T Technology > TK Electrical engineering. Electronics Nuclear engineering |
| Faculty/Division: | College of Engineering |
| Depositing User: | Mr. Nik Ahmad Nasyrun Nik Abd Malik |
| Date Deposited: | 08 Jan 2024 10:21 |
| Last Modified: | 08 Jan 2024 10:21 |
| URI: | http://umpir.ump.edu.my/id/eprint/39912 |
| Download Statistic: | View Download Statistics |
Actions (login required)
 |
View Item |