Hi, in my application I would need to rotate the image 180° but I found only [discussions and links](https://docs.luxonis.com/projects/sdk/en/latest/samples/CameraComponent/sdk_rotated/#camera-rotated-preview) that use depthai-sdk but my application uses normal depthai, with what code can I emulate the result? thanks. es (invented) camRgb.setRotation(180°)

Hi @"jakaskerl"#p22457, I tried the code you sent me, but I can't seem to fit it to my code, and in the end to turn the output around I simply reversed the image ( e.g. [::-1,::-1,] ), but by doing so I notice that the detection is no longer optimized as before, my code is[ based on this](https://github.com/luxonis/depthai-python/blob/main/examples/SpatialDetection/spatial_mobilenet.py), can you help me for a version that also turns the detection part around so that even if I put the camera upside down I always have the AI optimized for the right flow? thank you very much Translated with DeepL.com (free version)

Rotated Camera

Simone

Hi, in my application I would need to rotate the image 180° but I found only discussions and links that use depthai-sdk but my application uses normal depthai, with what code can I emulate the result? thanks.

es (invented) camRgb.setRotation(180°)

jakaskerl

Hi @Simone
Use a manip inbetween to achieve the same thing

manip = pipeline.create(dai.node.ImageManip)
manip.initialConfig.setResize(300, 300)
manip.initialConfig.setRotationDegrees(180)

Thanks,
Jaka

Simone

Hi jakaskerl, I tried the code you sent me, but I can't seem to fit it to my code, and in the end to turn the output around I simply reversed the image ( e.g. [::-1,::-1,] ), but by doing so I notice that the detection is no longer optimized as before, my code is based on this, can you help me for a version that also turns the detection part around so that even if I put the camera upside down I always have the AI optimized for the right flow?

thank you very much

Translated with DeepL.com (free version)

jakaskerl

Hi @Simone
Do you need frames rotated before the NN model or after. "Before" is usually used to correct the image rotation in case the camera is turned to the side.

For the first approach place manip node between camera node and the NN node.

If you want to only rotate the images after, you can do so on the host side (as you did with [::-1,::-1]), but you also have to change the X and Y of the detections bbox so they fit the new rotated image.
luxonis/depthai-pythonblob/main/examples/SpatialDetection/spatial_mobilenet.py#L143C13-L147C46

Thanks,
Jaka

Simone

jakaskerl yes I have already done the rotation also of the frame in the x and y coordinates but I am really talking about a recognition problem ( e.g. of people ) with the camera reversed and this rotation method the recognition rate drops dratically, so should I turn NN model with the code provided above?
manip = pipeline.create(dai.node.ImageManip)

manip.initialConfig.setResize(300, 300)

manip.initialConfig.setRotationDegrees(180)

jakaskerl

Simone the recognition rate drops dratically

That is expected yes. You need to rotate the image back to it's standard rotation before passing it to the NN node.

Simone self.camRgb.preview.link(self.manipRgb.inputImage)

You are linking the preview to image manip and rotating it, but you are not sending that rotated frame out to spatial detection network node:

Simone self.camRgb.preview.link(self.spatialDetectionNetwork.input)

The image is therefore not rotated.

Simone

    self.nnBlobPath = setting.model
    self.labelMap = ["person"]
    self.syncNN = True
    self.pipeline = dai.Pipeline()        
    self.camRgb = self.pipeline.create(dai.node.ColorCamera)
    self.camRgbRotate = self.pipeline.create(dai.node.ColorCamera)
    self.objectTracker = self.pipeline.create(dai.node.ObjectTracker)
    self.spatialDetectionNetwork = self.pipeline.create(dai.node.MobileNetSpatialDetectionNetwork)
    self.monoLeft = self.pipeline.create(dai.node.MonoCamera)
    self.monoRight = self.pipeline.create(dai.node.MonoCamera)
    self.stereo = self.pipeline.create(dai.node.StereoDepth)
    self.xoutRgb = self.pipeline.create(dai.node.XLinkOut)
    self.xoutNN = self.pipeline.create(dai.node.XLinkOut)
    self.xoutDepth = self.pipeline.create(dai.node.XLinkOut)
    self.objectTracker.setDetectionLabelsToTrack([15])  # track only person
    #possible tracking types: ZERO_TERM_COLOR_HISTOGRAM, ZERO_TERM_IMAGELESS, SHORT_TERM_IMAGELESS, SHORT_TERM_KCF
    self.objectTracker.setTrackerType(dai.TrackerType.ZERO_TERM_COLOR_HISTOGRAM)
    # take the smallest ID when new object is tracked, possible options: SMALLEST_ID, UNIQUE_ID
    self.objectTracker.setTrackerIdAssignmentPolicy(dai.TrackerIdAssignmentPolicy.SMALLEST_ID)

    # Setting Stream Name
    self.xoutRgb.setStreamName("rgb")
    self.xoutNN.setStreamName("detections")
    self.xoutDepth.setStreamName("depth")
    # Setting Properties
    self.camRgb.setPreviewSize(setting.x, setting.y)
    self.camRgb.setResolution(dai.ColorCameraProperties.SensorResolution.THE_1080_P)
    self.camRgb.setInterleaved(False)
    self.camRgb.setFps(40)
    self.camRgb.setColorOrder(dai.ColorCameraProperties.ColorOrder.BGR)
    # Setting stereo resolution
    self.monoLeft.setResolution(dai.MonoCameraProperties.SensorResolution.THE_400_P)
    self.monoLeft.setCamera("left")
    self.monoRight.setResolution(dai.MonoCameraProperties.SensorResolution.THE_400_P)
    self.monoRight.setCamera("right")

    self.manipRgb = self.pipeline.create(dai.node.ImageManip)
    self.manipRgb.initialConfig.setResize(setting.x, setting.y)
    self.manipRgb.initialConfig.setRotationDegrees(180)
    self.camRgb.preview.link(self.manipRgb.inputImage)

    self.manipRgbOut = self.pipeline.create(dai.node.XLinkOut)
    self.manipRgbOut.setStreamName("manip_rgb")
    self.manipRgb.out.link(self.manipRgbOut.input)

    
    #self.camRgb.setIspScale(1, 3) # You don't need to downscale (4k -> 720P) video frames
    #Squeeze the frame
    self.camRgb.setPreviewKeepAspectRatio(False)
    # Setting node configs
    self.stereo.initialConfig.setConfidenceThreshold(setting.sCT)
    # Align depth map to the perspective of RGB camera, on which inference is done
    self.stereo.setDepthAlign(dai.CameraBoardSocket.CAM_A)
    self.stereo.setSubpixel(False)
    self.stereo.setOutputSize(self.monoLeft.getResolutionWidth(), self.monoLeft.getResolutionHeight())
    # Setting parameters for Spatial Detection
    self.spatialDetectionNetwork.setBlobPath(self.nnBlobPath)
    self.spatialDetectionNetwork.setConfidenceThreshold(setting.CO)
    self.spatialDetectionNetwork.input.setBlocking(False)
    self.spatialDetectionNetwork.setBoundingBoxScaleFactor(0.9)
    self.spatialDetectionNetwork.setDepthLowerThreshold(setting.minDet)
    self.spatialDetectionNetwork.setDepthUpperThreshold(setting.maxDet) #distanza di riconoscimento
    # Linking
    self.monoLeft.out.link(self.stereo.left)
    self.monoRight.out.link(self.stereo.right)
    self.camRgb.preview.link(self.spatialDetectionNetwork.input)
    self.objectTracker.passthroughTrackerFrame.link(self.xoutRgb.input)
    self.objectTracker.out.link(self.xoutNN.input)

    if self.syncNN:
        self.spatialDetectionNetwork.passthrough.link(self.objectTracker.inputTrackerFrame)
    else:
        self.camRgb.preview.link(self.objectTracker.inputTrackerFrame)

    self.spatialDetectionNetwork.passthrough.link(self.objectTracker.inputDetectionFrame)
    self.spatialDetectionNetwork.out.link(self.objectTracker.inputDetections)        
    self.stereo.depth.link(self.spatialDetectionNetwork.inputDepth)`

this is the section of my code where I need to create the pipes for the camera, however I connect to multiple cameras at the same time so this section is common to multiple cams.
only afterwards during a cycle where I check to connect to all the required cameras I check whether they are to be shot or not and I had it set up like this

if setting.degreesCams[setting.ipCams.index(deviceInfo.name)] == 1: previewQueue = device.getOutputQueue(name="manip_rgb", maxSize=4, blocking=False) else: previewQueue = device.getOutputQueue(name="rgb", maxSize=4, blocking=False) detectionNNQueue = device.getOutputQueue(name="detections", maxSize=4, blocking=False) depthQueue = device.getOutputQueue(name="depth", maxSize=4, blocking=False)

manages to connect but after 5 previews it crashes with no output