🤖 Guide: Implementing Face Tracking for WALL-E

 

This guide outlines how to enable Face Tracking on your WALL-E robot using OpenCV. This feature allows WALL-E to detect a human face and automatically rotate its head to follow the movement.

 

1. Prerequisites

 

Ensure that the OpenCV data files (Haar Cascades) are installed on your Raspberry Pi.

Bash

sudo apt install opencv-data -y

 

2. Locate Data File

 

Find the absolute path of the haarcascade_frontalface_default.xml file, as the system might not locate it automatically.

Bash

find /usr -name "haarcascade_frontalface_default.xml"
# Common path: /usr/share/opencv4/haarcascades/haarcascade_frontalface_default.xml

 

3. Modify Backend (app.py)

 

Update the USBCameraStreamer class in web_interface/app.py to include the face detection and tracking logic.

 

A. Initialization (__init__)

 

Load the Haar Cascade model using the absolute path found in Step 2 and initialize the head position variable.

Python

def __init__(self):
    self.camera = cv2.VideoCapture(0)
    # Load the face detection model using the absolute path
    self.face_cascade = cv2.CascadeClassifier('/usr/share/opencv4/haarcascades/haarcascade_frontalface_default.xml')
    self.is_running = False
    self.head_pos = 50  # Initial head position (Center)

 

B. Tracking Logic (get_frame)

 

Implement the logic to detect faces and send servo commands to the Arduino.

Python

def get_frame(self):
    if self.is_running:
        success, frame = self.camera.read()
        if success:
            # 1. Convert to grayscale for performance
            gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
            
            # 2. Detect faces
            faces = self.face_cascade.detectMultiScale(gray, 1.1, 4)

            for (x, y, w, h) in faces:
                # Draw a rectangle around the face
                cv2.rectangle(frame, (x, y), (x+w, y+h), (255, 0, 0), 2)
                
                # 3. Calculate the center of the face
                x_center = x + (w / 2)
                
                # 4. Control Motors (Center is 320, Deadzone is 40px)
                global arduino
                
                # If face is on the left -> Rotate head left (Decrease angle)
                if x_center < 280: 
                    if self.head_pos > 0:
                        self.head_pos -= 1
                        arduino.send_command("G" + str(self.head_pos))
                        
                # If face is on the right -> Rotate head right (Increase angle)
                elif x_center > 360: 
                    if self.head_pos < 100:
                        self.head_pos += 1
                        arduino.send_command("G" + str(self.head_pos))

            ret, buffer = cv2.imencode('.jpg', frame)
            return buffer.tobytes()
    return None

 

4. Run & Test

 

Restart the WALL-E service to apply the changes.

Bash

sudo systemd restart walle.service
sudo systemd status walle.service

• Note: Verify that the service status is active (running).