Object, Age and Gender Detection Using OpenCV's Deep Neural Networks

This is a Python project that performs real-time age and gender detection on faces using OpenCV's deep neural networks. The script also performs object detection on the full-sized frame to detect objects within the face region.

The script uses pre-trained models, which are loaded into memory using the cv2.dnn.readNet method. These models include:

• age_net.caffemodel and age_deploy.prototxt for age detection
• caffemodel: age_net.caffemodel |caffemodel_url: https://github.com/GilLevi/AgeGenderDeepLearning/raw/master/models/age_net.caffemodel
• gender_net.caffemodel and gender_deploy.prototxt for gender detection
• caffemodel: gender_net.caffemodel | caffemodel_url: https://github.com/GilLevi/AgeGenderDeepLearning/raw/master/models/gender_net.caffemodel
• opencv_face_detector_uint8.pb and opencv_face_detector.pbtxt for face detection
• MobileNetSSD_deploy.caffemodel and MobileNetSSD_deploy.prototxt for object detection

The script captures frames from the default camera (index 0) using cv2.VideoCapture(0) and performs the following steps for each frame:

1. Resizes the frame to a smaller size for better optimization.
2. Performs face detection on the smaller frame using the face detection model.
3. If a face is detected, performs gender and age detection on the face region using their respective models.
4. Performs object detection on the full-sized frame using the object detection model within the face region.
5. Draws the results of the detections onto the original frame and displays them in two windows: one for age and gender detection and the other for object detection.
6. Prints the time taken to process each frame.

Requirements

To run this project, you need to have the following installed:

• Python 3.x
• OpenCV
• NumPy

You can install the required packages by running the following command:

pip install opencv-python numpy

Usage

To run the script, execute the following command in your terminal:

python age_gender_detection.py

This will launch the script and start capturing frames from your default camera.

Credits

This project was developed by Dev Jihad