How to use PySpin with opencv python?

Question

I just bought a FLIR BlackFlyS USB3.0 camera. I can grap frames from the camera but I am not able to use that frame with opencv without saving them first. Is there anyone who knows how to convert them to use in opencv?

I searched on the internet about everything that include "PySpin" word and found this book. I have tried to use PySpinCapture which is mentioned in this book but I couldn't figure it out anyway.

capture = PySpinCapture.PySpinCapture(0, roi=(0, 0, 960, 600),binningRadius=2,isMonochrome=True)

ret, frame = capture.read()

cv2.imshow("image",frame)

cv2.waitKey(0)

I expect the see the image but it throws an error

_PySpin.SpinnakerException: Spinnaker: GenICam::AccessException= Node is not writable. : AccessException thrown in node 'PixelFormat' while calling 'PixelFormat.SetIntValue()' (file 'EnumerationT.h', line 83) [-2006]
terminate called after throwing an instance of 'Spinnaker::Exception'

Answer 1

One year later, and not sure if my response will help, but I figured out that you can just get the RGB numpy array from a PySpin Image by using the GetData() function.

So you could do without the PySpinCapture module and just do something like the following.

import PySpin
import cv2

serial = '18475994' #Probably different for you although I also use a BlackFly USB3.0

system = PySpin.System.GetInstance()

blackFly_list = system.GetCameras()

blackFly = blackFly_list.GetBySerial(serial)

height = blackFly.Height()
width = blackFly.Width()
channels = 1

fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter('test_vid.avi',fourcc, blackFly.AcquisitionFrameRate(), (blackFly.Width(), blackFly.Height()), False) #The last argument should be True if you are recording in color.


blackFly.Init()
blackFly.AcquisitionMode.SetValue(PySpin.AcquisitionMode_Continuous)
blackFly.BeginAcquisition()

nFrames = 1000

for _ in range(nFrames):
    im = blackFly.GetNextImage()    

    im_cv2_format = im.GetData().reshape(height,width,channels)

    # Here I am writing the image to a Video, but once you could save the image as something and just do whatever you want with it.
    out.write(im_cv2_format)
    im.release()

out.release()    

In this code example I want to create an AVI video file with 1000 grabbed frames. The im.GetData() returns a 1-D numpy array which can then be converted into the correct dimensions with reshape. I have seen some talks about using the UMat class, but it does not seem to be necessary to make it work in this case. Perhaps it helps performance, but I am not sure :)