Fusion Identity - Ensemble Approach for Age Gender Detection
Shruti Dhumal, Omkar Awari, Shrutika Jadhav, Mohanish Khambadkar, Shubhangi Kale · 2025
Facial data has been always substantial when it comes to machine learning where it is crucial in facial recognition.The research exercises combination of CNN, Resnet 50 and Inception V3 algorithms as part of ensemble learning frame-work, applying the same to diverse dataset for training and validation. The past developments in fields of deep learning possess numerous challenges due to complexity , performance parameters, system limitations and robustness of models for age and gender detection systems. The balanced accuracy and significant performance is provided by integrating the models like CNN,Resnet 50, Inception V3. The predictions of each individual model is combined to have more accurate predictions when it comes to real time data. CNN has accuracy of 89.47, Inception v3 has 91.01 percent accuracy and Resnet 50 has resulted into accuracy of 91.08 percent and ensemble model to 95 percent. The proposed methodology of the research has more generalized approach for stable system for age and gender prediction along with improved accuracy and improving the real time predictions by integrating deep learning models through ensemble learning.