jakaskerl
I run the example you mentioned above, and it worked fine with MBNetSSD. However, When I switched it with YOLO it's giving errors:
1. First thing I changed is nnpath to blob file:
nnPathDefault = str((Path(__file__).parent / Path('model/mobilenet-ssd_openvino_2021.4_6shave.blob')).resolve().absolute())
to
nnPathDefault = str((Path(__file__).parent / Path('model/tiny-yolo-v4_openvino_2021.2_6shave.blob')).resolve().absolute())
and then it gave me error like this:
[DetectionNetwork(1)] [error] Mask is not defined for output layer with width '26'. Define at pipeline build time using: 'setAnchorMasks' for 'side26'.

2. Then I added this lines to the code:

anchorMasks = {
"side26" : [1,2,3],
"side13" : [3,4,5]
}
detectionNetwork.setAnchorMasks(anchorMasks)

After adding this lines now I'm getting this error:

with dai.Device(pipeline, device) as device:

[18443010C190BE0800] [192.168.1.234] [11.263] [system] [critical] Fatal error. Please report to developers. Log: 'Fatal error on MSS CPU: trap: 09, address: 8008EC3C' '0'

[18443010C190BE0800] [192.168.1.234] [1699258753.950] [host] [warning] Monitor thread (device: 18443010C190BE0800 [192.168.1.234]) - ping was missed, closing the device connection

Traceback (most recent call last):

File "/home/fakha/Work Projects/YoloDepthAI/gen2-yolo/device-decoding/adding_tracker.py", line 87, in <module>

imgFrame = preview.get()

RuntimeError: Communication exception - possible device error/misconfiguration. Original message 'Couldn't read data from stream: 'preview' (X_LINK_ERROR)'

Is there something that I'm doing wrong?

Thank you for your time,
Fakhrullo

    Hi Fakhrullo
    I just swapped the model to the yolo from https://docs.luxonis.com/projects/api/en/latest/samples/Yolo/tiny_yolo/:

    #!/usr/bin/env python3
    
    from pathlib import Path
    import cv2
    import depthai as dai
    import numpy as np
    import time
    import argparse
    
    # tiny yolo v4 label texts
    labelMap = [
        "person",         "bicycle",    "car",           "motorbike",     "aeroplane",   "bus",           "train",
        "truck",          "boat",       "traffic light", "fire hydrant",  "stop sign",   "parking meter", "bench",
        "bird",           "cat",        "dog",           "horse",         "sheep",       "cow",           "elephant",
        "bear",           "zebra",      "giraffe",       "backpack",      "umbrella",    "handbag",       "tie",
        "suitcase",       "frisbee",    "skis",          "snowboard",     "sports ball", "kite",          "baseball bat",
        "baseball glove", "skateboard", "surfboard",     "tennis racket", "bottle",      "wine glass",    "cup",
        "fork",           "knife",      "spoon",         "bowl",          "banana",      "apple",         "sandwich",
        "orange",         "broccoli",   "carrot",        "hot dog",       "pizza",       "donut",         "cake",
        "chair",          "sofa",       "pottedplant",   "bed",           "diningtable", "toilet",        "tvmonitor",
        "laptop",         "mouse",      "remote",        "keyboard",      "cell phone",  "microwave",     "oven",
        "toaster",        "sink",       "refrigerator",  "book",          "clock",       "vase",          "scissors",
        "teddy bear",     "hair drier", "toothbrush"
    ]
    
    nnPath = str((Path(__file__).parent / Path('../models/yolo-v4-tiny-tf_openvino_2021.4_6shave.blob')).resolve().absolute())
    
    
    
    # Create pipeline
    pipeline = dai.Pipeline()
    
    # Define sources and outputs
    camRgb = pipeline.create(dai.node.ColorCamera)
    detectionNetwork = pipeline.create(dai.node.YoloDetectionNetwork)
    objectTracker = pipeline.create(dai.node.ObjectTracker)
    
    xlinkOut = pipeline.create(dai.node.XLinkOut)
    trackerOut = pipeline.create(dai.node.XLinkOut)
    
    xlinkOut.setStreamName("preview")
    trackerOut.setStreamName("tracklets")
    
    # Properties
    camRgb.setPreviewSize(416, 416)
    camRgb.setResolution(dai.ColorCameraProperties.SensorResolution.THE_1080_P)
    camRgb.setInterleaved(False)
    camRgb.setColorOrder(dai.ColorCameraProperties.ColorOrder.BGR)
    camRgb.setFps(40)
    
    # testing MobileNet DetectionNetwork
    # Network specific settings
    detectionNetwork.setConfidenceThreshold(0.5)
    detectionNetwork.setNumClasses(80)
    detectionNetwork.setCoordinateSize(4)
    detectionNetwork.setAnchors([10, 14, 23, 27, 37, 58, 81, 82, 135, 169, 344, 319])
    detectionNetwork.setAnchorMasks({"side26": [1, 2, 3], "side13": [3, 4, 5]})
    detectionNetwork.setIouThreshold(0.5)
    detectionNetwork.setBlobPath(nnPath)
    detectionNetwork.setNumInferenceThreads(2)
    detectionNetwork.input.setBlocking(False)
    
    objectTracker.setDetectionLabelsToTrack([15])  # track only person
    # possible tracking types: ZERO_TERM_COLOR_HISTOGRAM, ZERO_TERM_IMAGELESS, SHORT_TERM_IMAGELESS, SHORT_TERM_KCF
    objectTracker.setTrackerType(dai.TrackerType.ZERO_TERM_COLOR_HISTOGRAM)
    # take the smallest ID when new object is tracked, possible options: SMALLEST_ID, UNIQUE_ID
    objectTracker.setTrackerIdAssignmentPolicy(dai.TrackerIdAssignmentPolicy.SMALLEST_ID)
    
    # Linking
    camRgb.preview.link(detectionNetwork.input)
    objectTracker.passthroughTrackerFrame.link(xlinkOut.input)
    
    
    detectionNetwork.passthrough.link(objectTracker.inputTrackerFrame)
    
    detectionNetwork.passthrough.link(objectTracker.inputDetectionFrame)
    detectionNetwork.out.link(objectTracker.inputDetections)
    objectTracker.out.link(trackerOut.input)
    
    # Connect to device and start pipeline
    with dai.Device(pipeline) as device:
    
        preview = device.getOutputQueue("preview", 4, False)
        tracklets = device.getOutputQueue("tracklets", 4, False)
    
        startTime = time.monotonic()
        counter = 0
        fps = 0
        frame = None
    
        while(True):
            imgFrame = preview.get()
            track = tracklets.get()
    
            counter+=1
            current_time = time.monotonic()
            if (current_time - startTime) > 1 :
                fps = counter / (current_time - startTime)
                counter = 0
                startTime = current_time
    
            color = (255, 0, 0)
            frame = imgFrame.getCvFrame()
            trackletsData = track.tracklets
            for t in trackletsData:
                roi = t.roi.denormalize(frame.shape[1], frame.shape[0])
                x1 = int(roi.topLeft().x)
                y1 = int(roi.topLeft().y)
                x2 = int(roi.bottomRight().x)
                y2 = int(roi.bottomRight().y)
    
                try:
                    label = labelMap[t.label]
                except:
                    label = t.label
    
                cv2.putText(frame, str(label), (x1 + 10, y1 + 20), cv2.FONT_HERSHEY_TRIPLEX, 0.5, 255)
                cv2.putText(frame, f"ID: {[t.id]}", (x1 + 10, y1 + 35), cv2.FONT_HERSHEY_TRIPLEX, 0.5, 255)
                cv2.putText(frame, t.status.name, (x1 + 10, y1 + 50), cv2.FONT_HERSHEY_TRIPLEX, 0.5, 255)
                cv2.rectangle(frame, (x1, y1), (x2, y2), color, cv2.FONT_HERSHEY_SIMPLEX)
    
            cv2.putText(frame, "NN fps: {:.2f}".format(fps), (2, frame.shape[0] - 4), cv2.FONT_HERSHEY_TRIPLEX, 0.4, color)
    
            cv2.imshow("tracker", frame)
    
            if cv2.waitKey(1) == ord('q'):
                break

    thanks,
    Jaka

      Hi jakaskerl
      Great work bro appriciate it so much 🙂

      Is it possible to put ROI before detection and tracking?

        Hi Fakhrullo
        Make a manip node with setCropRect() to crop to specified size of the preview stream and setResize() to make sure the input fits the NN first layer size.

        Thanks,
        Jaka

          Hi jakaskerl
          Thank you very much, I will continue working on this project if I have any questions I'll notify you

          Thanks,
          Fakhrullo

          Hi jakaskerl

          How are you doing bro? I have a question regarding that project, I adjusted the project as I want and it's working as expected, now I have to test detection and tracking with crowd, currently I can't put camera to street or any public places, so I wanted to test it with videos. However, I couldn't manage to integrate the video reading in the project, I tried using this method https://docs.luxonis.com/projects/api/en/latest/samples/ObjectTracker/object_tracker_video/#source-code
          In the beginnign I tried to add this to my code but I couldn't get it right, so I added Yolo to the code in the documentation I mentioned above. But it's not running on device, how can I test my project with video? can you help me with that please.

          Thanks,
          Fakhrullo

          Hi jakaskerl

          I added manip mode as you said, I added like this:
          # Creating Manip node
          manip = pipeline.create(dai.node.ImageManip)
          # Setting CropRect for the Region of Interest
          manip.initialConfig.setCropRect(*custom_roi)
          # Setting Resize for the neural network input size
          manip.initialConfig.setResize(640, 640)
          # Setting maximum output frame size based on the desired output dimensions
          max_output_width = 640
          max_output_height = 640
          max_output_frame_size = 3 * max_output_width * max_output_height # Assuming 3 channels for BGR image
          manip.setMaxOutputFrameSize(max_output_frame_size)
          # Connecting Manip node to ColorCamera
          camRgb.preview.link(manip.inputImage)
          # Connecting Manip node to YoloDetectionNetwork
          manip.out.link(detectionNetwork.input)

          But I'm getting black screen, Where am I making mistake? is there any tutorial or documentation how to do that?

          Hi @Fakhrullo
          Can I see the full code?
          If you are inputting the preview image into the manip, make sure it's larger than the output image, otherwise the crop makes no sense.
          Maybe you could link the .video output and set the frame type to RGB.

          Thanks,
          Jaka

            Hi jakaskerl
            Here's the code

            from pathlib import Path

            import cv2

            import depthai as dai

            import time

            from environs import Env
            env = Env()env.read_env()
            MxID = env('MxID')# Set your custom ROI coordinates (x, y, width, height)custom_roi = (350, 250, 640, 640) # Example coordinates, adjust as needed
            # tiny yolo v4 label textslabelMap = [ "person", "bicycle", "car", "motorbike", "aeroplane", "bus", "train", "truck", "boat", "traffic light", "fire hydrant", "stop sign", "parking meter", "bench", "bird", "cat", "dog", "horse", "sheep", "cow", "elephant", "bear", "zebra", "giraffe", "backpack", "umbrella", "handbag", "tie", "suitcase", "frisbee", "skis", "snowboard", "sports ball", "kite", "baseball bat", "baseball glove", "skateboard", "surfboard", "tennis racket", "bottle", "wine glass", "cup", "fork", "knife", "spoon", "bowl", "banana", "apple", "sandwich", "orange", "broccoli", "carrot", "hot dog", "pizza", "donut", "cake", "chair", "sofa", "pottedplant", "bed", "diningtable", "toilet", "tvmonitor", "laptop", "mouse", "remote", "keyboard", "cell phone", "microwave", "oven", "toaster", "sink", "refrigerator", "book", "clock", "vase", "scissors", "teddy bear", "hair drier", "toothbrush"]
            nnPath = str((Path(__file__).parent / Path('model/yolov6n_coco_640x640_openvino_2022.1_6shave.blob')).resolve().absolute())
            # Create pipelinepipeline = dai.Pipeline()
            # Define sources and outputscamRgb = pipeline.create(dai.node.ColorCamera)detectionNetwork = pipeline.create(dai.node.YoloDetectionNetwork)objectTracker = pipeline.create(dai.node.ObjectTracker)
            xlinkOut = pipeline.create(dai.node.XLinkOut)trackerOut = pipeline.create(dai.node.XLinkOut)
            xlinkOut.setStreamName("preview")trackerOut.setStreamName("tracklets")
            # Creating Manip nodemanip = pipeline.create(dai.node.ImageManip)# Setting CropRect for the Region of Interestmanip.initialConfig.setCropRect(*custom_roi)# Setting Resize for the neural network input sizemanip.initialConfig.setResize(640, 640)# Setting maximum output frame size based on the desired output dimensionsmax_output_width = 640max_output_height = 640max_output_frame_size = 3 * max_output_width * max_output_height # Assuming 3 channels for BGR imagemanip.setMaxOutputFrameSize(max_output_frame_size)
            # Propertiesif MxID == "14442C10C1AD3FD700": camRgb.setImageOrientation(dai.CameraImageOrientation.HORIZONTAL_MIRROR)camRgb.setPreviewSize(640, 640)camRgb.setResolution(dai.ColorCameraProperties.SensorResolution.THE_1080_P)camRgb.setInterleaved(False)camRgb.setColorOrder(dai.ColorCameraProperties.ColorOrder.BGR)camRgb.setFps(40)
            # Network specific settingsdetectionNetwork.setConfidenceThreshold(0.5)detectionNetwork.setNumClasses(80)detectionNetwork.setCoordinateSize(4)# detectionNetwork.setAnchors([10, 14, 23, 27, 37, 58, 81, 82, 135, 169, 344, 319]) #YOLOv4 uchun# detectionNetwork.setAnchorMasks({"side26": [1, 2, 3], "side13": [3, 4, 5]})detectionNetwork.setAnchors([10,13, 16,30, 33,23, 30,61, 62,45, 59,119, 116,90, 156,198, 373,326]) #YOLOv5 uchundetectionNetwork.setAnchorMasks({"side52": [0,1,2], "side26": [3,4,5], "side13": [6,7,8]})detectionNetwork.setIouThreshold(0.5)detectionNetwork.setBlobPath(nnPath)detectionNetwork.setNumInferenceThreads(2)detectionNetwork.input.setBlocking(False)
            objectTracker.setDetectionLabelsToTrack([0]) # track only person# possible tracking types: ZERO_TERM_COLOR_HISTOGRAM, ZERO_TERM_IMAGELESS, SHORT_TERM_IMAGELESS, SHORT_TERM_KCFobjectTracker.setTrackerType(dai.TrackerType.ZERO_TERM_COLOR_HISTOGRAM)# take the smallest ID when new object is tracked, possible options: SMALLEST_ID, UNIQUE_IDobjectTracker.setTrackerIdAssignmentPolicy(dai.TrackerIdAssignmentPolicy.SMALLEST_ID)
            #Linking# Connecting Manip node to ColorCameracamRgb.preview.link(manip.inputImage)# Connecting Manip node to YoloDetectionNetworkmanip.out.link(detectionNetwork.input)# camRgb.preview.link(detectionNetwork.input)objectTracker.passthroughTrackerFrame.link(xlinkOut.input)

            detectionNetwork.passthrough.link(objectTracker.inputTrackerFrame)
            detectionNetwork.passthrough.link(objectTracker.inputDetectionFrame)detectionNetwork.out.link(objectTracker.inputDetections)objectTracker.out.link(trackerOut.input)
            device = dai.DeviceInfo(MxID)
            # Connect to device and start pipelinewith dai.Device(pipeline, device) as device:
            preview = device.getOutputQueue("preview", 4, False) tracklets = device.getOutputQueue("tracklets", 4, False)
            startTime = time.monotonic() counter = 0 fps = 0 frame = None
            while(True): imgFrame = preview.get() track = tracklets.get()
            counter+=1 current_time = time.monotonic() if (current_time - startTime) > 1 : fps = counter / (current_time - startTime) counter = 0 startTime = current_time
            color = (255, 0, 0) text_color = (0, 0, 255) rectangle = (111, 147, 26)
            frame = imgFrame.getCvFrame() trackletsData = track.tracklets for t in trackletsData: if t.status.name == "TRACKED": roi = t.roi.denormalize(frame.shape[1], frame.shape[0]) x1 = int(roi.topLeft().x) y1 = int(roi.topLeft().y) x2 = int(roi.bottomRight().x) y2 = int(roi.bottomRight().y)
            try: label = labelMap[t.label] except: label = t.label # if t.status.name == 'TRACKED': cv2.putText(frame, str(label), (x1 + 10, y1 + 20), cv2.FONT_HERSHEY_TRIPLEX, 0.5, text_color) cv2.putText(frame, f"ID: {[t.id]}", (x1 + 10, y1 + 45), cv2.FONT_HERSHEY_TRIPLEX, 0.5, text_color) cv2.putText(frame, t.status.name, (x1 + 10, y1 + 70), cv2.FONT_HERSHEY_TRIPLEX, 0.5, text_color) cv2.rectangle(frame, (x1, y1), (x2, y2), rectangle, cv2.FONT_HERSHEY_SIMPLEX)
            cv2.putText(frame, "FPS: {:.2f}".format(fps), (2, frame.shape[0] - 4), cv2.FONT_HERSHEY_TRIPLEX, 0.6, text_color)
            cv2.imshow("tracker", frame)
            if cv2.waitKey(1) == ord('q'): break

            Here's better version:

            from pathlib import Path
            import cv2
            import depthai as dai
            import time
            
            from environs import Env
            
            env = Env()
            env.read_env()
            
            MxID = env('MxID')
            # Set your custom ROI coordinates (x, y, width, height)
            custom_roi = (350, 250, 640, 640)  # Example coordinates, adjust as needed
            
            # tiny yolo v4 label texts
            labelMap = [
                "person",         "bicycle",    "car",           "motorbike",     "aeroplane",   "bus",           "train",
                "truck",          "boat",       "traffic light", "fire hydrant",  "stop sign",   "parking meter", "bench",
                "bird",           "cat",        "dog",           "horse",         "sheep",       "cow",           "elephant",
                "bear",           "zebra",      "giraffe",       "backpack",      "umbrella",    "handbag",       "tie",
                "suitcase",       "frisbee",    "skis",          "snowboard",     "sports ball", "kite",          "baseball bat",
                "baseball glove", "skateboard", "surfboard",     "tennis racket", "bottle",      "wine glass",    "cup",
                "fork",           "knife",      "spoon",         "bowl",          "banana",      "apple",         "sandwich",
                "orange",         "broccoli",   "carrot",        "hot dog",       "pizza",       "donut",         "cake",
                "chair",          "sofa",       "pottedplant",   "bed",           "diningtable", "toilet",        "tvmonitor",
                "laptop",         "mouse",      "remote",        "keyboard",      "cell phone",  "microwave",     "oven",
                "toaster",        "sink",       "refrigerator",  "book",          "clock",       "vase",          "scissors",
                "teddy bear",     "hair drier", "toothbrush"
            ]
            
            nnPath = str((Path(__file__).parent / Path('model/yolov6n_coco_640x640_openvino_2022.1_6shave.blob')).resolve().absolute())
            
            # Create pipeline
            pipeline = dai.Pipeline()
            
            # Define sources and outputs
            camRgb = pipeline.create(dai.node.ColorCamera)
            detectionNetwork = pipeline.create(dai.node.YoloDetectionNetwork)
            objectTracker = pipeline.create(dai.node.ObjectTracker)
            
            xlinkOut = pipeline.create(dai.node.XLinkOut)
            trackerOut = pipeline.create(dai.node.XLinkOut)
            
            xlinkOut.setStreamName("preview")
            trackerOut.setStreamName("tracklets")
            
            # Creating Manip node
            manip = pipeline.create(dai.node.ImageManip)
            # Setting CropRect for the Region of Interest
            manip.initialConfig.setCropRect(*custom_roi)
            # Setting Resize for the neural network input size
            manip.initialConfig.setResize(640, 640)
            # Setting maximum output frame size based on the desired output dimensions
            max_output_width = 640
            max_output_height = 640
            max_output_frame_size = 3 * max_output_width * max_output_height # Assuming 3 channels for BGR image
            manip.setMaxOutputFrameSize(max_output_frame_size)
            
            # Properties
            if MxID == "14442C10C1AD3FD700":
                camRgb.setImageOrientation(dai.CameraImageOrientation.HORIZONTAL_MIRROR)
            camRgb.setPreviewSize(640, 640)
            camRgb.setResolution(dai.ColorCameraProperties.SensorResolution.THE_1080_P)
            camRgb.setInterleaved(False)
            camRgb.setColorOrder(dai.ColorCameraProperties.ColorOrder.BGR)
            camRgb.setFps(40)
            
            # Network specific settings
            detectionNetwork.setConfidenceThreshold(0.5)
            detectionNetwork.setNumClasses(80)
            detectionNetwork.setCoordinateSize(4)
            # detectionNetwork.setAnchors([10, 14, 23, 27, 37, 58, 81, 82, 135, 169, 344, 319]) #YOLOv4 uchun
            # detectionNetwork.setAnchorMasks({"side26": [1, 2, 3], "side13": [3, 4, 5]})
            detectionNetwork.setAnchors([10,13, 16,30, 33,23, 30,61, 62,45, 59,119, 116,90, 156,198, 373,326]) #YOLOv5 uchun
            detectionNetwork.setAnchorMasks({"side52": [0,1,2], "side26": [3,4,5], "side13": [6,7,8]})
            detectionNetwork.setIouThreshold(0.5)
            detectionNetwork.setBlobPath(nnPath)
            detectionNetwork.setNumInferenceThreads(2)
            detectionNetwork.input.setBlocking(False)
            
            objectTracker.setDetectionLabelsToTrack([0])  # track only person
            # possible tracking types: ZERO_TERM_COLOR_HISTOGRAM, ZERO_TERM_IMAGELESS, SHORT_TERM_IMAGELESS, SHORT_TERM_KCF
            objectTracker.setTrackerType(dai.TrackerType.ZERO_TERM_COLOR_HISTOGRAM)
            # take the smallest ID when new object is tracked, possible options: SMALLEST_ID, UNIQUE_ID
            objectTracker.setTrackerIdAssignmentPolicy(dai.TrackerIdAssignmentPolicy.SMALLEST_ID)
            
            #Linking
            # Connecting Manip node to ColorCamera
            camRgb.preview.link(manip.inputImage)
            # Connecting Manip node to YoloDetectionNetwork
            manip.out.link(detectionNetwork.input)
            # camRgb.preview.link(detectionNetwork.input)
            objectTracker.passthroughTrackerFrame.link(xlinkOut.input)
            
            
            detectionNetwork.passthrough.link(objectTracker.inputTrackerFrame)
            
            detectionNetwork.passthrough.link(objectTracker.inputDetectionFrame)
            detectionNetwork.out.link(objectTracker.inputDetections)
            objectTracker.out.link(trackerOut.input)
            
            device = dai.DeviceInfo(MxID)
            
            # Connect to device and start pipeline
            with dai.Device(pipeline, device) as device:
            
                preview = device.getOutputQueue("preview", 4, False)
                tracklets = device.getOutputQueue("tracklets", 4, False)
            
                startTime = time.monotonic()
                counter = 0
                fps = 0
                frame = None
            
                while(True):
                    imgFrame = preview.get()
                    track = tracklets.get()
            
                    counter+=1
                    current_time = time.monotonic()
                    if (current_time - startTime) > 1 :
                        fps = counter / (current_time - startTime)
                        counter = 0
                        startTime = current_time
            
                    color = (255, 0, 0)
                    text_color = (0, 0, 255)
                    rectangle = (111, 147, 26)
            
                    frame = imgFrame.getCvFrame()
                    trackletsData = track.tracklets
                    for t in trackletsData:
                        if t.status.name == "TRACKED":
                            roi = t.roi.denormalize(frame.shape[1], frame.shape[0])
                            x1 = int(roi.topLeft().x)
                            y1 = int(roi.topLeft().y)
                            x2 = int(roi.bottomRight().x)
                            y2 = int(roi.bottomRight().y)
            
                            try:
                                label = labelMap[t.label]
                            except:
                                label = t.label
                            # if t.status.name == 'TRACKED':
                            cv2.putText(frame, str(label), (x1 + 10, y1 + 20), cv2.FONT_HERSHEY_TRIPLEX, 0.5, text_color)
                            cv2.putText(frame, f"ID: {[t.id]}", (x1 + 10, y1 + 45), cv2.FONT_HERSHEY_TRIPLEX, 0.5, text_color)
                            cv2.putText(frame, t.status.name, (x1 + 10, y1 + 70), cv2.FONT_HERSHEY_TRIPLEX, 0.5, text_color)
                            cv2.rectangle(frame, (x1, y1), (x2, y2), rectangle, cv2.FONT_HERSHEY_SIMPLEX)
            
                    cv2.putText(frame, "FPS: {:.2f}".format(fps), (2, frame.shape[0] - 4), cv2.FONT_HERSHEY_TRIPLEX, 0.6, text_color)
            
                    cv2.imshow("tracker", frame)
            
                    if cv2.waitKey(1) == ord('q'):
                        break

              Hi Fakhrullo
              setCropRect() expects normalized values. Divide it by 416, then I believe it should work.

              Thanks,
              Jaka

                5 days later

                Hi jakaskerl
                Thanks it worked fine, I'm getting cropped frame.
                Is it possible to fuse cropped frame and whole frame? I mean it shows whole frame and cropped frame within whole frame may be with a box? If It is impossible how can I show whole and cropped frames in 2 different windows?

                  Hi Fakhrullo
                  Sorry, I don't really understand what you are trying to achieve. Both frames are numpy arrays which means you can easily stack one on top of the other.

                  Thanks,
                  Jaka

                    Hi jakaskerl
                    Yeah, basically I want to retrieve both full and cropped frames, to achieve that what should I do?

                    Fakhrullo camRgb.preview.link(manip.inputImage)

                    Here, you can pipe the preview stream to a new XLink node, so you can view it on host side.
                    This will enable you to see both the full preview frame as well as cropped frame (on which the inference was made).

                    Thanks,
                    Jaka