CECvT: Initial Diagnosis of Anomalies in Thermal Images

Given the global competitive landscape, it is imperative that businesses maintain and manage their facilities continuously to enhance efficiency productivity for sustaining competitiveness. Hence, a new hybrid model called contrast enhancement convolutional vision transformer (CECvT) was developed in this study enables fault diagnosis without physical contact with factory equipment ensure accurate initial detection risk of machine damage or interference. This leverages thermal imaging as an apt source early anomaly equipment. A module employing techniques integrated address edge information loss when utilizing images. Moreover, network performance enhanced by fusing advantages neural (CNN) Transformer models. Notably, design allows deriving detailed feature necessary diagnostics harnessing multiscale extract concatenate features. The proposed method’s evaluated using dataset provided AI Hub. When juxtaposed CNN, Transformer, CNN-Transformer models, demonstrated superior accuracy 96.17%. Furthermore, achieved most at inception abnormalities than other networks. thus has potential preferrable various thermal-imaging-based applications owing its excellent precision during diagnosis.