Classifying images into interpretative classes has grown increasingly challenging. Generally, the performance of a model with regard to accuracy is primarily influenced by the trained dataset and the chosen algorithm, among other factors. This study proposes a hybridization of deep convolutional neural network and machine learning classifiers for the purpose of classifying scenery imagery into different classes of building, forest, glacier, mountain, and sea. The rationale of the hybridization approach is that CNN has the capacity to reduce the dimensionality of feature vectors by transforming them into low-dimensional representations which makes subsequent machine learning classifiers more efficient and less prone to overfitting. In this research, the ResNet50 model was employed to extract features for classifiers such as Linear Support Vector Machine (LinearSVM), Gaussian Naive Bayes (GNB), eXtreme Gradient Boosting (XGBoost), Random Forest (RF), and Light Gradient Boost Machine (LGBM). The dataset used to evaluate the model was obtained from Kaggle. The LGBM-based model performed the best as it obtained the highest classification accuracy of 0.81, followed by XGBoost, RF, LinearSVM, a fully connected ResNet50, and GNB, with classification accuracies of 0.78, 0.78, 0.69, 0.68 and 0.64, respectively. The study showed that a fully connected ResNet50 model had an inferior performance on its own than when used in conjunction with other classifiers.