I've had some issues getting standalone mode running with getCvFrame running, running into the error that the 'lbp.ImgFrame' object has no attribute 'getCvFrame'
which, according to the docs, should be a functional attribute to query.
Here's the relevant code:
pipeline = dai.Pipeline()
camRgb = pipeline.create(dai.node.ColorCamera)
camRgb.setPreviewSize(300, 300)
camRgb.setFps(10)
camRgb.setInterleaved(False)
nn = pipeline.create(dai.node.MobileNetDetectionNetwork)
nn.setConfidenceThreshold(0.5)
nn.setBlobPath(blobconverter.from_zoo(name="mobilenet-ssd", shaves=6))
frame_builder = pipeline.create(dai.node.VideoEncoder)
camRgb.preview.link(nn.input)
frame_builder.setDefaultProfilePreset(camRgb.getFps(), dai.VideoEncoderProperties.Profile.MJPEG)
camRgb.video.link(frame_builder.input)
script = pipeline.create(dai.node.Script)
script.setProcessor(dai.ProcessorType.LEON_CSS)
nn.out.link(script.inputs['detections'])
frame_builder.bitstream.link(script.inputs['frame'])
and then in the Script node
imgFrame = node.io['frame'].get()
cvFrame = imgFrame.getCvFrame()
I'd presume I'm missing something straightforward. I'm eventually hoping to be able to implement a non-blocking queue of the last few frames once this is fixed, to skip lagging behind realtime on heavy operations.