Look this code:
import numpy as np
import cv2
class T(cv2.KeyPoint):
def __init__(self, pt):
super().__init__()
self.pt = pt
def calculate_corners(A):
A_gray = cv2.cvtColor(A, cv2.COLOR_BGR2GRAY)
pa = cv2.goodFeaturesToTrack(A_gray, maxCorners=100, qualityLevel=0.01, minDistance=15)
pa = np.squeeze(pa)
kpa = []
for coor in pa:
kpa.append(T(tuple(coor)))
return kpa
cap = cv2.VideoCapture("E:\\video\\test.mp4")
while True:
frame = cap.read()[1]
if frame is None:
break
kpa = calculate_corners(frame)
frame_corner = cv2.drawKeypoints(frame, kpa, outImage=None, color=(255, 0, 125))
cv2.imshow('frame_corner', frame_corner)
cv2.waitKey(1)
cv2.destroyAllWindows()
cap.release()
This code will crash in my system(Windows10, python3.7.3, opencv4.1.0)
After test, I'm sure this is caused by class T. I guess that class T dose not inherit release moudle of cv2.KeyPoint, so it cause memory leak. It's just my conjecture. And I didn't know how to fix it. Could anyone give me some advice? Thanks a lot!
↧