Reputation: 33
I just wrote this simple code to track circles, it is not tunned at all and I want to fix my error before I go on. It says:
ValueError: cannot reshape array of size 6 into shape (3,)
I tried looking online, but I'm in over my head. Can someone point me in the right direction?
import cv2
import numpy as np
cap = cv2.VideoCapture(0)
def circle_image(image, circles):
image_circles = np.zeros_like(image)
if circles is not None:
for circle in circles:
rad, x1, y1 = circle.reshape(3)
cv2.circle(image_circles,(x1,y1),rad,(255,0,0), 10)
return image_circles
while cap.isOpened():
ret, frame = cap.read()
if ret == True:
gray_frame = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
blur_frame = cv2.GaussianBlur(gray_frame, (5, 5), 0)
canny_frame = cv2.Canny(blur_frame, 100, 100)
tires = cv2.HoughCircles(canny_frame,cv2.HOUGH_GRADIENT, 1, 20, param1=50, param2=30, minRadius=0, maxRadius=0)
circles_to_blend = circle_image(canny_frame, tires)
combo_image = cv2.addWeighted(canny_frame, 0.8, circles_to_blend, 1, 1)
cv2.imshow('frame', combo_image)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
cap.release()
cv2.destroyAllWindows()
Upvotes: 0
Views: 132
Reputation: 11420
I actually had to test the HoughCircles in python to understand the problem. I don not know why, but the HoughCircles function returns in python a list of list of circles. So you want to do
for circle in circles[0,:]:
for it to work. However, in the documentation you can clearly see that this is wrong:
rad, x1, y1 = circle.reshape(3)
The circle should be x,y, rad and also you do not need to reshape. In some cases it may be 4 values (the votes) and this line will fail, since 4 values cannot be reshaped as 3, but you can do the following:
x1, y1, rad = circle[:3] # this takes the first 3 numbers
I think after these changes your code should work.
Upvotes: 3