FUEL & CHEMICAL PROCESSES ›› 2023, Vol. 54 ›› Issue (6): 42-44.
Previous Articles Next Articles
Online:
Published:
Abstract: Based on the data recorded of the battery basement,the YOLOv4-tiny single stage target detection algorithm is used to detect the status of the reversing machine and waste gas disc.Due to the slow reasoning speed of YOLOv4-tiny,the real-time requirements can not be achieved,in that case,the YOLOv4-tiny algorithm shall be improved.The Rep VGG network framework is used to replace the original backbone feature extraction network CSPDarknet53 framework,and the Rep-YOLOv4-tiny algorithm is obtained.The size of the model is greatly reduced to only 13 343 KB,which reduces the hardware requirements and makes it easier to deploy on the hardware platform.The reasoning speed is increased after optimization,the detection speed reaches 24.76 frames/s,and the average accuracy reaches 99.88%,which realizes the real-time detection of the waste gas disc state and meets the specific requirements of plant.
Key words: Object detection, Single stage, Waste gas disc state, Model optimization
CLC Number:
TQ520.5
Sun Shengming Wu Qiuling Wu Jianming. Research on state detection algorithm of waste gas disc based on improved YOLOv4-tiny#br#[J]. FUEL & CHEMICAL PROCESSES, 2023, 54(6): 42-44.
0 / / Recommend
Add to citation manager EndNote|Reference Manager|ProCite|BibTeX|RefWorks
URL: https://journal03.magtechjournal.com/Jwk3_rlyhg/EN/
https://journal03.magtechjournal.com/Jwk3_rlyhg/EN/Y2023/V54/I6/42