IJET 2024 Vol.16(3): 139-142
DOI: 10.7763/IJET.2024.V16.1270
Truck Dumper Control System with Safety Unit Using Real-Time Deep Learning
Nittaya Kerdprasop1,*, Apirak Worrakantapon1, Paradee Chuaybamroong2, and Kittisak Kerdprasop1
1. School of Computer Engineering, Suranaree University of Technology, Thailand
2. Department of Environmental Science, Thammasat University, Thailand
Email: nittaya@sut.ac.th (N.K.); apirak.wor@gmail.com (A.W.); paradee@tu.ac.th (P.C.); kerdpras@sut.ac.th (K.K.)
*Corresponding author
Manuscript received December 30, 2023; revised March 4, 2024; accepted April 15, 2024; published August 9, 2024
Abstract—Presently, industrial factories have moved toward the era of intelligent automation systems, especially, animal feed industries. They contained several automation machines such as a mixer, pellet mill, and truck dumper. The truck dumper has been used to lift the whole truck and dump the raw material into an intake hopper. However, during its process, neighbor areas are identified as a risk area. Thus, if there is a person standing on the truck dumper platform at the time, this may cause fatal injury or death. In this study, a new proposal to develop a model of automatic human detection using a convolutional neural network that learns and recognizes the important characteristics of the target objects has been introduced. The objective of the study was to develop a system of human detection and risk-level identification in the truck dumper control system to prevent accidents. Herein, several experiments have been conducted in order to select the effective and optimal object detection architecture to apply to the truck dumper control system. In the results, the YOLOv4 model outperformed the Faster Region-Convolution Neural Network (R-CNN) model in both precision and processing speed. The average precision was 99.93% on a day-time dataset and 94.25% on a night-time dataset. The overall accuracy of risk identification was 94.18%. An average processing speed was 31.96 frames per second.
Keywords—deep learning, You Only Look Once (YOLO), truck dumper control system, human detection
Cite: Nittaya Kerdprasop, Apirak Worrakantapon, Paradee Chuaybamroong, and Kittisak Kerdprasop, "Truck Dumper Control System with Safety Unit Using Real-Time Deep Learning," International Journal of Engineering and Technology, vol. 16, no. 3, pp. 139-142, 2024.
