# Digital Zoom Crops and displays the largest detected person from the full‑resolution RGB stream, and updates AE/AF regions to match the crop. Uses a YOLO model from the Model Zoo for person detection and remaps detections to the full‑resolution source. ## Demo This example requires the DepthAI v3 API, see [installation instructions](https://docs.luxonis.com/software-v3/depthai.md). ## Pipeline ### examples/digital_zoom.pipeline.json ```json { "pipeline": { "connections": [ { "node1Id": 0, "node1Output": "0", "node1OutputGroup": "dynamicOutputs", "node2Id": 6, "node2Input": "inputImage", "node2InputGroup": "" }, { "node1Id": 0, "node1Output": "0", "node1OutputGroup": "dynamicOutputs", "node2Id": 2, "node2Input": "inputImage", "node2InputGroup": "" }, { "node1Id": 2, "node1Output": "out", "node1OutputGroup": "", "node2Id": 4, "node2Input": "in", "node2InputGroup": "" }, { "node1Id": 4, "node1Output": "out", "node1OutputGroup": "", "node2Id": 5, "node2Input": "in", "node2InputGroup": "" }, { "node1Id": 4, "node1Output": "passthrough", "node1OutputGroup": "", "node2Id": 12, "node2Input": "in", "node2InputGroup": "" }, { "node1Id": 5, "node1Output": "out", "node1OutputGroup": "", "node2Id": 14, "node2Input": "in", "node2InputGroup": "" }, { "node1Id": 6, "node1Output": "out", "node1OutputGroup": "", "node2Id": 16, "node2Input": "in", "node2InputGroup": "" }, { "node1Id": 9, "node1Output": "out", "node1OutputGroup": "", "node2Id": 0, "node2Input": "inputControl", "node2InputGroup": "" }, { "node1Id": 11, "node1Output": "out", "node1OutputGroup": "", "node2Id": 6, "node2Input": "inputConfig", "node2InputGroup": "" } ], "globalProperties": { "calibData": null, "cameraTuningBlobSize": null, "cameraTuningBlobUri": "", "eepromId": 0, "leonCssFrequencyHz": 700000000.0, "leonMssFrequencyHz": 700000000.0, "pipelineName": null, "pipelineVersion": null, "sippBufferSize": 18432, "sippDmaBufferSize": 16384, "xlinkChunkSize": -1 }, "nodes": [ [ 0, { "alias": "", "id": 0, "ioInfo": [ [ [ "", "inputControl" ], { "blocking": true, "group": "", "id": 0, "name": "inputControl", "queueSize": 3, "type": 3, "waitForMessage": false } ], [ [ "dynamicOutputs", "0" ], { "blocking": false, "group": "dynamicOutputs", "id": 3, "name": "0", "queueSize": 8, "type": 0, "waitForMessage": false } ], [ [ "", "mockIsp" ], { "blocking": true, "group": "", "id": 1, "name": "mockIsp", "queueSize": 8, "type": 3, "waitForMessage": false } ], [ [ "", "raw" ], { "blocking": false, "group": "", "id": 2, "name": "raw", "queueSize": 8, "type": 0, "waitForMessage": false } ] ], "logLevel": 3, "name": "Camera", "parentId": -1, "properties": { "boardSocket": 0, "cameraName": "", "fps": -1.0, "imageOrientation": -1, "initialControl": { "aeLockMode": false, "aeMaxExposureTimeUs": 0, "aeRegion": { "height": 0, "priority": 0, "width": 0, "x": 0, "y": 0 }, "afRegion": { "height": 0, "priority": 0, "width": 0, "x": 0, "y": 0 }, "antiBandingMode": 0, "autoFocusMode": 3, "awbLockMode": false, "awbMode": 0, "brightness": 0, "captureIntent": 0, "chromaDenoise": 0, "cmdMask": 0, "contrast": 0, "controlMode": 0, "effectMode": 0, "enableHdr": false, "expCompensation": 0, "expManual": { "exposureTimeUs": 0, "frameDurationUs": 0, "sensitivityIso": 0 }, "frameSyncMode": 0, "lensPosAutoInfinity": 0, "lensPosAutoMacro": 0, "lensPosition": 0, "lensPositionRaw": 0.0, "lowPowerNumFramesBurst": 0, "lowPowerNumFramesDiscard": 0, "lumaDenoise": 0, "miscControls": [], "saturation": 0, "sceneMode": 0, "sharpness": 0, "strobeConfig": { "activeLevel": 0, "enable": 0, "gpioNumber": 0 }, "strobeTimings": { "durationUs": 0, "exposureBeginOffsetUs": 0, "exposureEndOffsetUs": 0 }, "wbColorTemp": 0 }, "isp3aFps": 0, "mockIspHeight": -1, "mockIspWidth": -1, "numFramesPoolIsp": 3, "numFramesPoolPreview": 4, "numFramesPoolRaw": 3, "numFramesPoolStill": 4, "numFramesPoolVideo": 4, "outputRequests": [ { "enableUndistortion": null, "fps": { "value": null }, "resizeMode": 0, "size": { "value": { "index": 0, "value": [ 4056, 3040 ] } }, "type": null } ], "resolutionHeight": -1, "resolutionWidth": -1, "sensorType": -1 } } ], [ 2, { "alias": "", "id": 2, "ioInfo": [ [ [ "", "inputConfig" ], { "blocking": true, "group": "", "id": 4, "name": "inputConfig", "queueSize": 3, "type": 3, "waitForMessage": false } ], [ [ "", "inputImage" ], { "blocking": true, "group": "", "id": 5, "name": "inputImage", "queueSize": 3, "type": 3, "waitForMessage": false } ], [ [ "", "out" ], { "blocking": false, "group": "", "id": 6, "name": "out", "queueSize": 8, "type": 0, "waitForMessage": false } ] ], "logLevel": 3, "name": "ImageManip", "parentId": -1, "properties": { "backend": 0, "initialConfig": { "base": { "background": 0, "backgroundB": 0, "backgroundG": 0, "backgroundR": 0, "center": true, "colormap": 0, "operations": [], "outputHeight": 384, "outputWidth": 512, "resizeMode": 1, "undistort": false }, "outputFrameType": 8, "reusePreviousImage": false, "skipCurrentImage": false }, "numFramesPool": 4, "outputFrameSize": 1048576, "performanceMode": 0 } } ], [ 12, { "alias": "", "id": 12, "ioInfo": [ [ [ "", "in" ], { "blocking": true, "group": "", "id": 17, "name": "in", "queueSize": 3, "type": 3, "waitForMessage": false } ] ], "logLevel": 3, "name": "XLinkOut", "parentId": -1, "properties": { "maxFpsLimit": -1.0, "metadataOnly": false, "streamName": "__x_4_passthrough" } } ], [ 4, { "alias": "neuralNetwork", "id": 4, "ioInfo": [ [ [ "", "in" ], { "blocking": true, "group": "", "id": 7, "name": "in", "queueSize": 3, "type": 3, "waitForMessage": true } ], [ [ "", "out" ], { "blocking": false, "group": "", "id": 8, "name": "out", "queueSize": 8, "type": 0, "waitForMessage": false } ], [ [ "", "passthrough" ], { "blocking": false, "group": "", "id": 9, "name": "passthrough", "queueSize": 8, "type": 0, "waitForMessage": false } ] ], "logLevel": 3, "name": "NeuralNetwork", "parentId": 3, "properties": { "backend": "", "backendProperties": {}, "blobSize": 8685208, "blobUri": "asset:__blob", "modelSource": 0, "modelUri": "", "numFrames": 8, "numNCEPerThread": 0, "numShavesPerThread": 0, "numThreads": 0 } } ], [ 5, { "alias": "detectionParser", "id": 5, "ioInfo": [ [ [ "", "in" ], { "blocking": true, "group": "", "id": 10, "name": "in", "queueSize": 1, "type": 3, "waitForMessage": true } ], [ [ "", "out" ], { "blocking": false, "group": "", "id": 11, "name": "out", "queueSize": 8, "type": 0, "waitForMessage": false } ] ], "logLevel": 3, "name": "DetectionParser", "parentId": 3, "properties": { "networkInputs": { "images": { "dataType": 1, "dims": [ 512, 384, 3, 1 ], "name": "images", "numDimensions": 4, "offset": 0, "order": 17185, "qpScale": 1.0, "qpZp": 0.0, "quantization": false, "strides": [] } }, "numFramesPool": 8, "parser": { "anchorMasks": {}, "anchors": [], "anchorsV2": [], "classNames": [ "person", "bicycle", "car", "motorcycle", "airplane", "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", "couch", "potted plant", "bed", "dining table", "toilet", "tv", "laptop", "mouse", "remote", "keyboard", "cell phone", "microwave", "oven", "toaster", "sink", "refrigerator", "book", "clock", "vase", "scissors", "teddy bear", "hair drier", "toothbrush" ], "classes": 80, "confidenceThreshold": 0.5, "coordinates": 4, "iouThreshold": 0.5, "nnFamily": 0, "subtype": "yolov6r2" } } } ], [ 6, { "alias": "", "id": 6, "ioInfo": [ [ [ "", "inputConfig" ], { "blocking": true, "group": "", "id": 12, "name": "inputConfig", "queueSize": 3, "type": 3, "waitForMessage": false } ], [ [ "", "inputImage" ], { "blocking": false, "group": "", "id": 13, "name": "inputImage", "queueSize": 1, "type": 3, "waitForMessage": false } ], [ [ "", "out" ], { "blocking": false, "group": "", "id": 14, "name": "out", "queueSize": 8, "type": 0, "waitForMessage": false } ] ], "logLevel": 3, "name": "ImageManip", "parentId": -1, "properties": { "backend": 0, "initialConfig": { "base": { "background": 0, "backgroundB": 0, "backgroundG": 0, "backgroundR": 0, "center": true, "colormap": 0, "operations": [], "outputHeight": 800, "outputWidth": 800, "resizeMode": 1, "undistort": false }, "outputFrameType": 33, "reusePreviousImage": false, "skipCurrentImage": false }, "numFramesPool": 4, "outputFrameSize": 1500300, "performanceMode": 0 } } ], [ 9, { "alias": "", "id": 9, "ioInfo": [ [ [ "", "out" ], { "blocking": false, "group": "", "id": 15, "name": "out", "queueSize": 8, "type": 0, "waitForMessage": false } ] ], "logLevel": 3, "name": "XLinkIn", "parentId": -1, "properties": { "maxDataSize": 18482688, "numFrames": 8, "streamName": "__x_0__inputControl" } } ], [ 11, { "alias": "", "id": 11, "ioInfo": [ [ [ "", "out" ], { "blocking": false, "group": "", "id": 16, "name": "out", "queueSize": 8, "type": 0, "waitForMessage": false } ] ], "logLevel": 3, "name": "XLinkIn", "parentId": -1, "properties": { "maxDataSize": 18482688, "numFrames": 8, "streamName": "__x_6__inputConfig" } } ], [ 14, { "alias": "", "id": 14, "ioInfo": [ [ [ "", "in" ], { "blocking": true, "group": "", "id": 18, "name": "in", "queueSize": 3, "type": 3, "waitForMessage": false } ] ], "logLevel": 3, "name": "XLinkOut", "parentId": -1, "properties": { "maxFpsLimit": -1.0, "metadataOnly": false, "streamName": "__x_5_out" } } ], [ 16, { "alias": "", "id": 16, "ioInfo": [ [ [ "", "in" ], { "blocking": true, "group": "", "id": 19, "name": "in", "queueSize": 3, "type": 3, "waitForMessage": false } ] ], "logLevel": 3, "name": "XLinkOut", "parentId": -1, "properties": { "maxFpsLimit": -1.0, "metadataOnly": false, "streamName": "__x_6_out" } } ] ] } } ``` ## Source code #### Python ```python #!/usr/bin/env python3 import cv2 import depthai as dai import time # This example shows how to crop out the biggest person from the maximum resolution of the camera and setting the autofocus and autoexposure region to the same ROI. def getBiggestPerson(imgDetections: dai.ImgDetections): biggestDetection = None biggestDetectionSize = 0 for detection in imgDetections.detections: if detection.label == 0: # Person size = (detection.xmax - detection.xmin) * (detection.ymax - detection.ymin) if size > biggestDetectionSize: biggestDetection = detection biggestDetectionSize = size return biggestDetection def displayFrame(name: str, frame: dai.ImgFrame, imgDetections: dai.ImgDetections, labelMap: dict): color = (0, 255, 0) assert imgDetections.getTransformation() is not None cvFrame = frame.getFrame() if frame.getType() == dai.ImgFrame.Type.RAW16 else frame.getCvFrame() for detection in imgDetections.detections: # Get the shape of the frame from which the detections originated for denormalization normShape = imgDetections.getTransformation().getSize() # Create rotated rectangle to remap rotRect = dai.RotatedRect(dai.Rect(dai.Point2f(detection.xmin, detection.ymin), dai.Point2f(detection.xmax, detection.ymax)).denormalize(normShape[0], normShape[1]), 0) # Remap the detection rectangle to target frame remapped = imgDetections.getTransformation().remapRectTo(frame.getTransformation(), rotRect) # Remapped rectangle could be rotated, so we get the bounding box bbox = [int(l) for l in remapped.getOuterRect()] cv2.putText( cvFrame, labelMap[detection.label], (bbox[0] + 10, bbox[1] + 20), cv2.FONT_HERSHEY_TRIPLEX, 0.5, 255, ) cv2.putText( cvFrame, f"{int(detection.confidence * 100)}%", (bbox[0] + 10, bbox[1] + 40), cv2.FONT_HERSHEY_TRIPLEX, 0.5, 255, ) cv2.rectangle(cvFrame, (bbox[0], bbox[1]), (bbox[2], bbox[3]), color, 2) # Show the frame cv2.imshow(name, cvFrame) def transformDetectionToSource(imgDetections: dai.ImgDetections, detection: dai.ImgDetection): normShape = imgDetections.getTransformation().getSize() sourceSize = imgDetections.getTransformation().getSourceSize() width, height = sourceSize rotRect = dai.RotatedRect(dai.Rect(dai.Point2f(detection.xmin, detection.ymin), dai.Point2f(detection.xmax, detection.ymax)).denormalize(normShape[0], normShape[1]), 0) rotRect = imgDetections.getTransformation().invTransformRect(rotRect) outerRect = rotRect.getOuterRect() firstPoint = dai.Point2f(max(0, min(outerRect[0], width)), max(0, min(outerRect[1], height))) secondPoint = dai.Point2f(max(0, min(outerRect[2], width)), max(0, min(outerRect[3], height))) return dai.Rect(firstPoint, secondPoint) device = dai.Device() modelPath = dai.getModelFromZoo(dai.NNModelDescription("yolov6-nano", platform=device.getPlatformAsString())) modelArchive = dai.NNArchive(modelPath) inputSize = modelArchive.getInputSize() type = modelArchive.getConfig().model.inputs[0].preprocessing.daiType if type: try: frameType = getattr(dai.ImgFrame.Type, type) except AttributeError: type = None if not type: if device.getPlatform() == dai.Platform.RVC2: frameType = dai.ImgFrame.Type.BGR888p else: frameType = dai.ImgFrame.Type.BGR888i # Create pipeline with dai.Pipeline(device) as pipeline: # Define source and output cam = pipeline.create(dai.node.Camera).build() cameraControlQueue = cam.inputControl.createInputQueue() fullResStream = cam.requestFullResolutionOutput(useHighestResolution=True) imageManip = pipeline.create(dai.node.ImageManip) imageManip.initialConfig.setOutputSize(inputSize[0], inputSize[1]) imageManip.initialConfig.setFrameType(frameType) fullResStream.link(imageManip.inputImage) detectionNetwork = pipeline.create(dai.node.DetectionNetwork).build(imageManip.out, modelArchive) labelMap = detectionNetwork.getClasses() imageManipCropOut = pipeline.create(dai.node.ImageManip) imageManipCropOut.setMaxOutputFrameSize(round(1000*1000*1.5)+300) imageManipCropOut.initialConfig.setOutputSize(800, 800) imageManipCropOut.inputImage.setBlocking(False) imageManipCropOut.inputImage.setMaxSize(1) imageManipConfigQueue = imageManipCropOut.inputConfig.createInputQueue() imageManipCropOutQueue = imageManipCropOut.out.createOutputQueue() fullResStream.link(imageManipCropOut.inputImage) videoQueue = detectionNetwork.passthrough.createOutputQueue() detectionQueue = detectionNetwork.out.createOutputQueue() # Connect to device and start pipeline pipeline.start() lastTimeToAutoFocus = time.time() while pipeline.isRunning(): videoIn = videoQueue.get() detections = detectionQueue.get() biggestDetection = getBiggestPerson(detections) if biggestDetection: sourceRect = transformDetectionToSource(detections, biggestDetection) configQueue = dai.ImageManipConfig() configQueue.addCrop(sourceRect, False) configQueue.setOutputSize(800, 800, dai.ImageManipConfig.ResizeMode.LETTERBOX) imageManipConfigQueue.send(configQueue) configCamera = dai.CameraControl() configCamera.setAutoExposureRegion(int(sourceRect.x), int(sourceRect.y), int(sourceRect.width), int(sourceRect.height)) if(time.time() - lastTimeToAutoFocus > 5): lastTimeToAutoFocus = time.time() configCamera.setAutoFocusRegion(int(sourceRect.x), int(sourceRect.y), int(sourceRect.width), int(sourceRect.height)) cameraControlQueue.send(configCamera) imageManipCropOutFrame = imageManipCropOutQueue.tryGet() if imageManipCropOutFrame is not None: assert isinstance(imageManipCropOutFrame, dai.ImgFrame) cv2.imshow("Cropped out frame", imageManipCropOutFrame.getCvFrame()) assert isinstance(videoIn, dai.ImgFrame) assert isinstance(detections, dai.ImgDetections) displayFrame("Full view video", videoIn, detections, labelMap) key = cv2.waitKey(1) if key == ord("q"): break ``` ### Need assistance? Head over to [Discussion Forum](https://discuss.luxonis.com/) for technical support or any other questions you might have.