The primary and most prevalent form of road deterioration involves the occurrence of cracks, posing a potential risk to the safety of roads and highways. Recognizing the elevated costs and potential inaccuracies associated with manual inspection methods, there has been a recent exploration of automated inspection solutions. In response to the challenges posed by manual methods, various machine learning techniques have been devised to address the high cost and potential errors in road safety assessments. This study introduces an automatic road classification system utilizing a hybrid combination of machine learning algorithms, specifically integrating Principal Component Analysis (PCA) and Convolutional Neural Network (CNN). RDD2020 dataset, involving PCA for feature extraction and CNN is utilized for road class classification. Both qualitative and quantitative analyses are conducted to assess the proposed methodology. Results demonstrate the efficacy of the approach, revealing elevated accuracy, recall, precision, and F-score parameters. These metrics collectively underscore the model's proficiency in accurately classifying diverse road types. A detailed comparative analysis with the Naïve Bayes algorithm further establishes the superior performance of the proposed method. This research contributes to the advancement of automatic road classification, emphasizing the practical viability of the hybrid model in enhancing classification accuracy. The findings provide valuable insights for the development of intelligent transportation systems and urban planning initiatives. The successful integration of PCA and CNN in road classification tasks showcases a promising avenue for addressing the complexity of diverse road environment.