Basketball Tracking OpenCV

The topic I’ll be dealing with this time is OpenCV Contour and tracking.

Result :

Let’s Start !!

# Important library
import math
import cv2
import cvzone
from cvzone.ColorModule import ColorFinder
import numpy as np


"""
pip install opencv-python
pip install cvzone
"""

This time, I will implement object tracking using only Opencv without using AI.

First, to detect a basketball, we’ll find the optimal HSV values using the code below.

import cv2
import cvzone
from cvzone.ColorModule import ColorFinder

# Create the color Finder object
# True : debug mode
myColorFinder = ColorFinder(True)
hsvVals = {'hmin': 8, 'smin': 138, 'vmin': 0, 'hmax': 142, 'smax': 255, 'vmax': 255}

# Variables
posListX, posListY = [], []
xList = [item for item in range(0, 1300)]
prediction = False

while True:
    # Grab the image
    img = cv2.imread("ball.png")
    img = img[0:900, :] # crop the image

    # Display
    img = cv2.resize(img, (0, 0), None, 0.7, 0.7)
    
    # Process the image using the ColorFinder update method
    imgColor, mask = myColorFinder.update(img, hsvVals)
    
    cv2.imshow("Image with Color", imgColor)

    # Break the loop if the 'q' key is pressed
    if cv2.waitKey(1) & 0xFF == ord('q'):
        break

cv2.destroyAllWindows()

Tracking :

import math
import cv2
import cvzone
from cvzone.ColorModule import ColorFinder
import numpy as np

# Initialize the Video
cap = cv2.VideoCapture('./Videos/vid (4).mp4')

# Create the color Finder object
myColorFinder = ColorFinder(False)
hsvVals = {'hmin': 8, 'smin': 138, 'vmin': 0, 'hmax': 142, 'smax': 255, 'vmax': 255}

# Variables
posListX, posListY = [], []
xList = [item for item in range(0, 1300)] # image shpae of x -> 1300x1080
prediction = False

while True:
    
    success, img = cap.read()
    img = img[0:900, :]
    
    # Display
    img = cv2.resize(img, (0, 0), None, 0.7, 0.7)
    
    # Process the image using the ColorFinder update method
    imgColor, mask = myColorFinder.update(img, hsvVals)
    
    imgContours, contours = cvzone.findContours(img, mask, minArea=500)
    
    if contours:
        posListX.append(contours[0]['center'][0])
        posListY.append(contours[0]['center'][1])
        
    if posListX and posListY: 
        # polunomial regression y = Ax ^ 2 + Bx + C
        A, B, C = np.polyfit(posListX, posListY, 2)
        
        
        for i, (posX, posY) in enumerate(zip(posListX, posListY)):
            pos = (posX, posY)
            cv2.circle(imgContours, pos, 5, (0, 255, 0), cv2.FILLED)
            if i >= 1:
                cv2.line(imgContours, pos, (posListX[i - 1], posListY[i - 1]), (0, 255, 0), 2)
        
        for x in xList:
            y = int(A * x**2 + B * x + C)
            cv2.circle(imgContours, (x, y), 2, (255, 0, 255), cv2.FILLED)
            
        print(posListX)
        if len(posListX) < 10 :
            # Prediction
            a = A
            b = B
            c = C - 590

            x = int((-b - math.sqrt(b ** 2 - (4 * a * c))) / (2 * a))
            # print(x)
            prediction = 100 < x < 150

        if prediction:
            cvzone.putTextRect(imgContours, "Basket", (50, 100),
                               scale=3, thickness=3, colorR=(0, 200, 0), offset=20)
        else:
            cvzone.putTextRect(imgContours, "No Basket", (50, 100),
                               scale=3, thickness=3, colorR=(0, 0, 200), offset=20)
            
    cv2.imshow("Image with countours", imgContours)
    
    # Break the loop if the 'q' key is pressed
    if cv2.waitKey(85) & 0xFF == ord('q'):
        break

# Release the video capture and close all windows
cap.release()
cv2.destroyAllWindows()

Important part :

The aspect we need to pay attention to is the area value of the contour.

# Find location of the Ball
imgContours, contours = cvzone.findContours(img, mask, minArea=500)

We use a polynomial to obtain a line predicting the trajectory of the basketball.

Here, since it’s 2-dimensional, we only calculate A, B, and C for the equation y = Ax² + Bx + C using np.polyfit.

A, B, C = np.polyfit(posListX, posListY, 2)

https://media.geeksforgeeks.org/wp-content/uploads/20230628110338/Polynomial-Functions-4-min.png

We use the quadratic formula to determine whether the basketball has entered the goal or not.

That’s all !

Summary :

1. We obtain the HSV values, extract contour values larger than 500

2. Save the center points of those contours, use a polynomial to determine the trajectory of the basketball

3. Check whether it has entered the goal or not using the quadratic formula.

The content I’m sharing this time is based on the video from the YouTuber below.

Link :

Legend OpenCV YouTuber