<p>The rapid proliferation of electronic devices has resulted in a significant growth in the generation of electronic waste. Improper disposal of these wastes creates serious environmental problems and also causes loss of valuable materials. Due to this, an efficient methodology is needed for identifying different types of electronic waste. In this work, a novel Faster Region Coordinate Gradient-based Random Puffer Fish model is proposed for accurate detection and classification of electronic waste. The Coordinate Faster Region-based Convolutional Neural Network is designed to detect and classify electronic waste that utilizes a coordinate attention mechanism for improving the spatial feature learning. It allows the model to detect small and irregular electronic waste objects more accurately. In addition, a Gradient-based Random Puffer Fish Optimization algorithm is introduced to fine-tune the hyperparameters of the Coordinate Faster Region-based Convolutional Neural Network. The hyperparameter tuning model leverages the advantages of puffer fish optimization and random update strategy, where the random update strategy improves the exploration ability and prevents the optimization process from getting trapped in local optima problems, thereby the proposed model achieves better convergence stability and improved detection performance. Experimental validation demonstrates that the Faster Region Coordinate Gradient-based Random Puffer Fish approach outperforms existing methods by achieving 98.91% accuracy, 0.981 Matthews Correlation Coefficient, 98.83% precision, and 98.64% F1-Score, which confirms that the proposed framework provides a superior and reliable solution for intelligent electronic waste classification and waste management.</p> Graphical abstract <p></p>