'opencv.imshow will cause jupyter notebook crash
I check other question on google or stackoverflow, they are talking about run cv2.imshow in script, but my code run in jupyter notebook.
Here is my configuration:
ubuntu 16.4x64
python 3.5
opencv 3.1.0
I start a jupyter notebook: here is the code I put it notebook:
%pylab notebook
import cv2
cvim2disp = cv2.imread('data/home.jpg')
cv2.imshow('HelloWorld', cvim2disp)
cv2.waitKey() #image will not show until this is called
cv2.destroyWindow('HelloWorld') #make sure window closes cleanly
When I execute these code. image will show in a pop up window, but I can not close this window by clicking the x on the top right corner, and a moment later, system will prompt me that the window is not responding, it will give me 2 choices: "wait" , "fore quit". if I hit wait, then It will show the same prompt later, If I hit 'fore quit', then the jupyter notebook kernel die and I have to start over.
I google around, many solution suggest that I should add this code
cv2.startWindowThread()
before imshow
, but situation get worse, the kernel hang forever!.
anybody have some idea what's going on.
Solution 1:[1]
%matplotlib inline
#The line above is necesary to show Matplotlib's plots inside a Jupyter Notebook
import cv2
from matplotlib import pyplot as plt
#Import image
image = cv2.imread("input_path")
#Show the image with matplotlib
plt.imshow(image)
plt.show()
Solution 2:[2]
I was having a similar problem, and could not come to a good solution with cv2.imshow()
in the Jupyter Notebook. I followed this stackoverflow answer, just using matplotlib to display the image.
import matplotlib.pyplot as plt
# load image using cv2....and do processing.
plt.imshow(cv2.cvtColor(image, cv2.COLOR_BGR2RGB))
# as opencv loads in BGR format by default, we want to show it in RGB.
plt.show()
Solution 3:[3]
The API documentation for cv2.waitKey() notes the following:
This function is the only method in HighGUI that can fetch and handle events, so it needs to be called periodically for normal event processing unless HighGUI is used within an environment that takes care of event processing.
So perhaps calling the function in an endless loop would make the window responsive? I haven't tested this, but maybe you would like to try the following:
import cv2
cvim2disp = cv2.imread('data/home.jpg')
cv2.imshow('img', cvim2disp)
while(True):
k = cv2.waitKey(33)
if k == -1: # if no key was pressed, -1 is returned
continue
else:
break
cv2.destroyWindow('img')
Solution 4:[4]
This will help you understand what is happening:
import cv2
cvim2disp = cv2.imread('data/home.jpg')
cv2.imshow('HelloWorld', cvim2disp)
cv2.waitKey(0)
cv2.destroyWindow('HelloWorld')
waitKey(0)
method is waiting for an input infinitely. When you see a frame of the corresponding image, do not try to close the image using close in top right corner.
Instead press some key. waitkey
method will take that as an input and it will return back a value. Further you can also check which key was pressed to close the frame.
Additionally waitKey(33)
will keep the frame active for 33 ms and then close it automatically.
destroyWindow()
will destroy the current frame if there.
destroyAllWindows()
will destroy all the frames currently present.
This will solve.
Solution 5:[5]
if your facing problem in google collab ,you can use this patch
from google.colab.patches import cv2_imshow
cv2_imshow(img)
Solution 6:[6]
%matplotlib inline
from matplotlib import pyplot as plt
img = cv2.imread(valid_img_paths[1])
plt.imshow(cv2.cvtColor(img, cv2.COLOR_BGR2RGB))
plt.show()
Solution 7:[7]
The following code works fine in Jupyter to show one image
%matplotlib inline
import cv2
from matplotlib import pyplot as plt
cap = cv2.VideoCapture(videoFName)
ret, image = cap.read()
image=cv2.resize(image,None,fx=0.25,fy=0.25,interpolation=cv2.INTER_AREA)
plt.imshow(image)
plt.show()
If you want to show the video instead of an image in a separate window, use the following code:
import cv2
cap = cv2.VideoCapture(videoFName)
while cap.isOpened():
ret, image = cap.read()
image=cv2.resize(image,None,fx=0.25,fy=0.25,interpolation=cv2.INTER_AREA)
cv2.imshow('image',image)
k = cv2.waitKey(30) & 0xff # press ESC to exit
if k == 27 or cv2.getWindowProperty('image', 0)<0:
break
cv2.destroyAllWindows()
cap.release()
Make sure the window name match, otherwise it will not work. In this case I use 'image' as window name.
Solution 8:[8]
The new window that opens up from Jupyter uses the same kernel as notebook. Just add this below to the code and it would work fine.
cv2.waitKey(0)
cv2.destroyAllWindows()
Solution 9:[9]
image = cv2.imread(file_path)
while True:
# Press 'q' for exit
exit_key = ord('q')
if cv2.waitKey(exit_key) & 255 == exit_key:
cv2.destroyAllWindows()
break
cv2.imshow('Image_title', image)
Solution 10:[10]
I have just developed a library that is exactly similar to cv2.imshow
and it can be used in both Jupyter
and colab
. It can update the window. Therefore you can easily see the video inside it. It uses HTML canvas and is browser friendly :)
Installation:
pip install opencv_jupyter_ui
Usage:
This is the replacement of cv2.imshow
for Jupiter. you need only to replace cv2.imshow
with cv2_imshow
. It will work in Jupiter.
First:
please import the library
from opencv_jupyter_ui import cv2_imshow
Then:
change cv2.imshow
->cv2_imshow
More details exist on the Github Repository. Working example
Solution 11:[11]
I am not sure if you can open a window from Jupyter Notebook. cv2.imshow expects a waitKey which doesn't work in Jupyter.
Here is what I have done (using OpenCV 3.3):
from IPython.display import display, HTML
import cv2
import base64
def imshow(name, imageArray):
_, png = cv2.imencode('.png', imageArray)
encoded = base64.b64encode(png)
return HTML(data='''<img alt="{0}" src="data:image/png;base64, {1}"/>'''.format(name, encoded.decode('ascii')))
img = cv2.imread('./media/baboon.jpg',cv2.IMREAD_COLOR)
imshow('baboon', img)
If you don't need to use cv2, just:
from IPython.display import Image
Image('./media/baboon.jpg')
Sources
This article follows the attribution requirements of Stack Overflow and is licensed under CC BY-SA 3.0.
Source: Stack Overflow