Age estimation from facial images using deep learning methods

This study focused on predicting age from facial images using DenseNet architectures and analyzed the Indian Movie Face Database (IMFDB). Age estimation is vital in sectors like security, healthcare, and marketing. Different DenseNet architectures, including DenseNet-121, DenseNet-169, DenseNet-201, and DenseNet-264, were trained and evaluated based on metrics such as accuracy, precision, recall, and F1 score. The dataset consisted of 19,906 facial images labeled with age categories. DenseNet-264 achieved the highest performance with 89.8% accuracy, 0.90 precision, 0.90 recall, and 0.90 F1 score. DenseNet-201 followed with 89.2% accuracy, while DenseNet-169 demonstrated a balanced generalization with 87.1% accuracy. DenseNet-121 provided a faster and computationally efficient alternative with 85.3% accuracy. ROC analysis revealed strong classification capabilities across all models, with minor generalization losses observed in the deepest model, DenseNet-264. In conclusion, DenseNet architectures proved to be effective solutions for age prediction. Model selection can be tailored to the application's requirements and resources. While DenseNet-264 offered the highest accuracy, DenseNet-121 served as a faster alternative with lower computational costs.