smsalaken commented 5 years ago

Hi, I have a use for pymba to track object on opencv. This needs to work similar to cv.VideoCapture() and allow the user to select an object using selectROI() method. I couldn't figure out how to get it done using pymba. Could you please give me some hint or example code to start with? Thanks

SunnyAVT commented 5 years ago

cv.VideoCapture() cannot be used to access Pymba camera. You can refer to attached code for how to access the camera, get image and set ROI (Camera_Set_ROI function as below). More samples can refer to https://github.com/morefigs/pymba/tree/master/examples/camera

from pymba import Vimba from pymba import VimbaException import cv2 import sys

def Camera_Set_ROI(camera, W: int, H: int) -> None: image_Height = camera.feature("Height") image_Width = camera.feature("Width") image_Height.value = H image_Width.value = W

def init_cameras():

vimba object

    vimba = Vimba()
    # Start vimba system
    vimba.startup()
    vmFactory = vimba.camera_ids()
    # Get connected cameras
    cams = [vimba.camera(id) for id in vmFactory]
    if len(cams) == 0:
        raise OSError("No camera present.")
    for idx, device in enumerate(vmFactory):
        print("Device {} ID: {}".format(idx, device))
    # Open cameras and print information
    for idx, cam in enumerate(cams):
        try:
            cam.open()
            Camera_Set_ROI(cam, 640, 480)
            cam.arm('SingleFrame')
        except VimbaException as e:
            if e.error_code == VimbaException.ERR_TIMEOUT:
                print(e)
                cam.disarm()
                cam.arm('SingleFrame')
            elif e.error_code == VimbaException.ERR_DEVICE_NOT_OPENED:
                print(e)
                cam.open()
                cam.arm('SingleFrame')
    return cams

def main(): PIXEL_FORMATS_CONVERSIONS = {'BayerRG8': cv2.COLOR_BAYER_RG2RGB} cams = init_cameras() flag = True while flag: try: for idx, cam in enumerate(cams): frame = cam.acquire_frame() img = frame.buffer_data_numpy()

convert color space if desired

            if(frame.pixel_format == 'BayerRG8'):
                img = cv2.cvtColor(img, PIXEL_FORMATS_CONVERSIONS[frame.pixel_format])
            img = cv2.resize(img, (400, 300))
            cv2.imshow("Cam {} - {}".format(idx, ["cam{}". format(idx)]), img)
            c = cv2.waitKey(5)
            if c == ord('q'):
                flag = False
    except KeyboardInterrupt:
        sys.exit(1)
for cam in cams:
    cam.disarm()
    cam.close()
cv2.destroyAllWindows()
Vimba().shutdown()

if name == 'main': main()

kumbhat941 commented 6 days ago

I am trying to set roi using the given program but with singleframe the image is completely dark, in my view may be due to lack of time to adjustment. So I tried using continuous mode for it but it does not work. Can you suggest me a better approach ?

morefigs / pymba

Using pymba with opencv trackers #103

vimba object

convert color space if desired