learning-coding-by-AI-help

After i learned several coursed by https://learn.deeplearning.ai/. I start to write some code with the ai chatbot help. here is some.

import cv2 import numpy as np

Load the video file

video = cv2.VideoCapture(r'your vide.mp4')

Get the frame rate of the video

fps = video.get(cv2.CAP_PROP_FPS)

Initialize previous coordinates, time, and speed flag

prev_x, prev_y = None, None prev_time = None speed_calculated = False

Real-world tennis court length and its length in video (in pixels)

court_length_meters = 23.77 court_length_pixels = 800 # replace this with your actual measurement

Calculate pixels to meters conversion factor

pixels_to_meters = court_length_meters / court_length_pixels

Define the speed threshold in meters per second to identify when the ball leaves the racket

speed_threshold = 10 # Adjust this value based on your observations

while True: ret, frame = video.read()

if not ret:
    break

# Convert the frame to grayscale (optional, depends on your detection strategy)
gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)

# Simple static thresholding as a placeholder for proper object detection
_, thresh = cv2.threshold(gray, 127, 255, cv2.THRESH_BINARY)

# Find contours (outlines of objects) in the thresholded image
contours, _ = cv2.findContours(thresh, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)

# Let's assume the largest contour in the frame is the ball 
if contours:
    largest_contour = max(contours, key=cv2.contourArea)
    x, y, w, h = cv2.boundingRect(largest_contour)
    cv2.rectangle(frame, (x, y), (x+w, y+h), (0, 255, 0), 2)
    
    # Get current time in seconds
    current_time = video.get(cv2.CAP_PROP_POS_MSEC) / 1000.0  # Convert milliseconds to seconds
    
    if prev_x is not None and prev_y is not None and prev_time is not None:
        # Calculate distance (Euclidean distance in pixels)
        distance_pixels = np.sqrt((x - prev_x)**2 + (y - prev_y)**2)
        
        # Convert distance to meters
        distance_meters = distance_pixels * pixels_to_meters
        
        # Calculate elapsed time
        elapsed_time = current_time - prev_time
        
        if elapsed_time > 0:  # Avoid division by zero
            # Calculate speed (meters per second)
            speed_mps = distance_meters / elapsed_time
            
            if speed_mps > speed_threshold and not speed_calculated:
                # Convert speed to kilometers per hour
                speed_kmph = speed_mps * 3.6
                print(f'Speed at racket leaving: {speed_kmph:.2f} km/h')
                speed_calculated = True
    
    # Update previous coordinates and time
    prev_x, prev_y = x, y
    prev_time = current_time

# Show the frame with the detected object
cv2.imshow('Frame', frame)

# Press 'q' to quit
if cv2.waitKey(1) & 0xFF == ord('q'):
    break

video.release() cv2.destroyAllWindows()