Resize neural network to 640x400

anapaulaSb

Hello, i am working on a code that gets the depth of an object. im using mobilenet-ssd.blob as the neural network. i want the size to be 640x400, however when i run it with that configuration i get this warning:

[warning] Input image (640x400) does not match NN (300x300)

how can i resize the NN to match the 640x400? i have tried with:

spatialDetectionNetwork.initialConfig.setResize(640, 400)

but its not working

this is the part of the code with the configuration i have:

imageManip.initialConfig.setResize(640, 400)

imageManip.initialConfig.setFrameType(dai.ImgFrame.Type.BGR888p)

monoLeft.setResolution(dai.MonoCameraProperties.SensorResolution.THE_400_P)

monoLeft.setCamera("left")

monoRight.setResolution(dai.MonoCameraProperties.SensorResolution.THE_400_P)

monoRight.setCamera("right")

stereo.setDefaultProfilePreset(dai.node.StereoDepth.PresetMode.HIGH_DENSITY)

stereo.setSubpixel(True)

camera_resolution = monoLeft.getResolution()

spatialDetectionNetwork.setConfidenceThreshold(0.5)

spatialDetectionNetwork.setBlobPath(nnPath)

spatialDetectionNetwork.input.setBlocking(False)

spatialDetectionNetwork.setBoundingBoxScaleFactor(0.5)

spatialDetectionNetwork.setDepthLowerThreshold(100)

spatialDetectionNetwork.setDepthUpperThreshold(5000)

monoLeft.out.link(stereo.left)

monoRight.out.link(stereo.right)

imageManip.out.link(spatialDetectionNetwork.input)

thanks!!

jakaskerl

Hi @anapaulaSb
The NN input shape depends on the model. Unless you change the model itself, this size is unchangeable.

You can downscale your camera ouput to from 640x400 to 300x300 using something like setPreviewSize() or an ImageManip node.
Choose a different model that has 640x400 input shape and convert it to .blob.

Thanks,
Jaka