The main goal of the project is to estimate age of a person based on image. During the course of development we have implemented gender detection model, although its evaluation was abandoned due to insatisfactory results. In the project UTKFace datasets are used for model training. (https://susanqq.github.io/UTKFace/)

Different approaches:

1. Convolutional Neural Networks
2. Transfer Learning with ResNet architecture
3. Transfer Learning with MobileNet architecture

The application itself may process the face of the person in real time by analyzing the camera input of the user, an image or a video. All use cases are present and selectable in the GUI.

face_rec_project/

dataset_descriptive_statistics.ipynb

/cnn/utk_age_cnn.ipynb

/transfer_learning/Tranfer_learning_models_and_testing.ipynb

/machine_learning_report.pdf