Solid waste management is important for environmental sustainability. Correct classification of recyclable materials plays a critical role in increasing the efficiency of recycling processes. In our research, hyperparameters of the EfficientNet model were optimized with Grey Wolf Optimizer (GWO) to increase this efficiency. In addition to hyperparameter optimization, the performance of machine learning algorithms such as Naive Bayes, Logistic Regression, and Multilayer Perceptron (MLP) was evaluated. Furthermore, the effect of these algorithms on the classification of features extracted from different deep learning models, including EfficientNet, MobileNet, and VGG, was investigated. The EfficientNet model optimized with GWO achieved the best performance with a 95.43% accuracy rate. These results showed that hyperparameter optimization applied to deep learning models increased the success in the solid waste classification problem. The integration of deep learning-based feature extraction and the optimization ability of GWO increased the classification performance while making the training process more efficient. Furthermore, proper hyperparameter tweaking enhanced the model’s overall performance by preventing overfitting. To sum up, this study shows how deep learning and optimization techniques work well in the waste management industry. The findings show that these technologies can provide significant improvements in recycling processes. Moreover, these methods have the potential to contribute to more efficient and sustainable management of recycling processes in real-world applications.