A CNN-based Prediction Model for Age, Gender, and Ethnicity Using Facial Images

A computational model can examine multiple critical demographic factors at once, including age, gender, and ethnicity. This is useful when cognitive robots need to execute multiple computer vision tasks at once and GPU and memory resources are limited. Multi-task learning allows a solitary model to proficiently handle multiple tasks simultaneously to do several classification and regression tasks by mastering single low-level representation in an effective method for doing this. An assumed relationship between age, gender, and ethnicity is taken care of using the combined estimation approach. In the current study, gender, age, ethnicity, and mood are predicted from facial photographs in real-time user profiling using a multi-task convolutional neural network. Three different architectures were looked at to see which one struck the optimum balance between accuracy and speed. The multitasking CNNs outperformed equivalent single-task CNNs in accuracy, execution speed, and memory use. This approach is particularly suitable for overcoming the processing resource constraints particular to such applications, and it can be used in real-world situations like embedded systems and smart cameras.