Object Detection Tutorial with OpenCV

Introduction

• Tutorial focuses on object detection using OpenCV.
• The goal is to have the computer verbally identify objects (e.g., apple, orange, cell phone).

Dependencies Installation

1. OpenCV Contrib Python

   • Command: pip install OpenCV-contrib-python
   • Contains extra libraries beyond basic OpenCV.

2. CLIB

   • Command: pip install cvlib
   • Used for object detection.

3. Text-to-Speech Libraries

   • Command: pip install gtts play sound
   • Used for verbal outputs.

4. Pi Object C

   • Command: pip3 install pyobjC
   • Enhances sound playback efficiency.

Importing Libraries

• Import necessary libraries:

  import cv2  
  import cvlib as cv  
  from cvlib.object_detection import draw_box  
  from gtts import gTTS  
  from playsound import playsound
  

Accessing Camera

• Use cv2.VideoCapture to access the camera.
• The index for the camera may vary (most use 0, but this tutorial uses 1).

Object Detection Loop

1. Retrieve frames from the video feed:

   ret, frame = video.read()
   

2. Detect objects with the following:

   boxes, labels, conf = cv.detect_common_objects(frame)
   

3. Draw boxes around detected objects:

   output_image = draw_box(frame, boxes, labels, conf)
   

4. Show the output image:

   cv2.imshow('Object Detection', output_image)
   

5. Exit condition with key press (e.g., 'Q').

Handling Detected Labels

• Create a list to store unique labels:

  labels = []
  

• Add labels to the list if they are not already present.
• Print the list of labels.

Constructing and Displaying Speech

1. Create a function for speech:

   def speech(text):
       print(text)
       # Set language and save audio.
   

2. Use string interpolation to format detected labels into natural language:

   • For example: "I found a person, an apple, a tie..."
   • Use lists and the join function for formatting.

3. Save audio output as MP3 file and play it:

   output.save('./sounds/output.mp3')
   playsound('./sounds/output.mp3')
   

Conclusion

• The program successfully detects objects and provides verbal feedback.
• Encouragement for further questions and feedback in comments.
• Reminder to like and subscribe for more tutorials.