Parameters Correlation of Deep Learning Model for Visual Inspection in the Automotive Industry

Welding seams' visual inspection is manually operated by humans in different fields, which makes the result of inspection subjective and expensive. At present, the integration of deep learning methods for welds classification is a research focus in engineering applications. This work intends to apprehend and emphasize the contribution of deep learning model explainability and the improvement of welding seams classification accuracy: two of the main metrics affecting production lines and cost in the automotive industry. This paper introduces a novel hybrid method that relies on combining the model prediction scores and visual explanation heatmap. The results reaches higher accuracy than a traditional deep learning model, by at least 18,2%. New perspectives in explaining and interpreting deep learning models are presented in this contribution.