Brtdku

I'm working on a school-project.

I've been given the task to identify objects when seeing them with a camera.

Video

Image

I've chosen to try template matching:

from matplotlib import pyplot as plt

import cv2



template = cv2.imread('./frog.jpg') # read image as template

cap = cv2.VideoCapture('./frog.mov') # simulate camera input



height = template.shape[0]

width = template.shape[1]



while True:

    ret, frame = cap.read() # read the current frame



    if ret is False:

        break

    res = cv2.matchTemplate(frame, template, cv2.TM_CCORR_NORMED)



    min_val, max_val, min_loc, max_loc = cv2.minMaxLoc(res)



    top_left = max_loc

    bottom_right = (top_left[0] + width, top_left[1] + height)



    cv2.rectangle(frame,top_left, bottom_right, 255, 2)



    match_found = False



    for i in res:

        if i.any() > 0.9999:

            print ('match found')

            match_found = True

            break 



    if match_found:

        plt.subplot(121),plt.imshow(res,cmap = 'gray')

        plt.title('Matching Result'), plt.xticks(), plt.yticks()

        plt.subplot(122),plt.imshow(frame,cmap = 'gray')

        plt.title('Detected Point'), plt.xticks(), plt.yticks()

        plt.suptitle('TM_CCORR_NORMED')

        plt.show()

The problem is, that almost everything is matching.
How can I get this to work ?

If you would use frog and bear images you will get 100% objects matching. Following script detects these images on video and shows names of detected objects:

from matplotlib import pyplot as plt

import cv2



images = {'Frog': cv2.imread('frog.jpg'), 'Bear': cv2.imread('bear.jpg')}

cap = cv2.VideoCapture('frog.mov')  # simulate camera input



while cap.isOpened():

    ret, frame = cap.read()  # read the current frame

    if ret:

        for name, image in images.items():

            match = cv2.matchTemplate(frame, image, cv2.TM_CCOEFF_NORMED)

            _, quality, _, max_loc = cv2.minMaxLoc(match)

            if quality > 0.99:



                top_left = max_loc

                bottom_right = (top_left[0] + image.shape[1], top_left[1] + image.shape[0])

                cv2.rectangle(frame, top_left, bottom_right, 255, 2)

                plt.subplot(121), plt.imshow(match, cmap='gray')

                plt.title('Matching Result'), plt.xticks(), plt.yticks()

                plt.subplot(122), plt.imshow(frame, cmap='gray')

                plt.title('Detected Point'), plt.xticks(), plt.yticks()

                plt.suptitle(name)

                plt.show()

    else:

        break

So what I have on the end: