Enhancing GUI Testing: Exploring Convolutional Neural Networks with Self-Learning Mechanisms for Automated UI Validation in Mobile Applications

Abstract

The suggested methodology provides effective UI layout categorization with systematic workflow starting from an input source gathering UI images. Data preprocessing improves data quality through normalization and imputation of missing values. Feature extraction is accomplished using deep learning-based detection of significant UI elements followed by the prediction phase classifying layouts as defective or non-defective. The automated method enhances UI validation with a minimum of human interaction and a maximum of software development. The performance of the model is confirmed by critical metrics and has high accuracy (95.65%), precision (96.80%), recall (94.50%), and an F1 score of 95.90%, showing it is good for detecting defects.