# RGB Encoding & Mono with MobilenetSSD & Depth

This example shows how to configure the depthai video encoder in h.265 format to encode the RGB camera input at Full-HD resolution
at 30FPS, and transfers the encoded video over XLINK to the host, saving it to disk as a video file. At the same time, a
MobileNetv2SSD network is ran on the frames from right grayscale camera, while the application also displays the depth map
produced by both of the grayscale cameras. Note that disparity is used in this case, as it colorizes in a more intuitive way.

Pressing Ctrl+C will stop the recording and then convert it using ffmpeg into an mp4 to make it playable. Note that ffmpeg will
need to be installed and runnable for the conversion to mp4 to succeed.

Be careful, this example saves encoded video to your host storage. So if you leave it running, you could fill up your storage on
your host.

It's a combination of [RGB Encoding](https://docs.luxonis.com/software/depthai/examples/rgb_encoding.md) and [Mono & MobilenetSSD
& Depth](https://docs.luxonis.com/software/depthai/examples/mono_depth_mobilenetssd.md).

### Similar samples:

 * [RGB Encoding](https://docs.luxonis.com/software/depthai/examples/rgb_encoding.md)
 * [RGB & Mono Encoding](https://docs.luxonis.com/software/depthai/examples/rgb_mono_encoding.md)
 * [Encoding Max Limit](https://docs.luxonis.com/software/depthai/examples/encoding_max_limit.md)
 * [RGB Encoding & MobilenetSSD](https://docs.luxonis.com/software/depthai/examples/rgb_encoding_mobilenet.md)
 * [RGB Encoding & Mono & MobilenetSSD](https://docs.luxonis.com/software/depthai/examples/rgb_encoding_mono_mobilenet.md)

## Demo

## Setup

Please run the [install script](https://github.com/luxonis/depthai-python/blob/main/examples/install_requirements.py) to download
all required dependencies. Please note that this script must be ran from git context, so you have to download the
[depthai-python](https://github.com/luxonis/depthai-python) repository first and then run the script

```bash
git clone https://github.com/luxonis/depthai-python.git
cd depthai-python/examples
python3 install_requirements.py
```

For additional information, please follow the [installation guide](https://docs.luxonis.com/software/depthai/manual-install.md).

## Source code

#### Python

```python
#!/usr/bin/env python3

from pathlib import Path
import sys
import cv2
import depthai as dai
import numpy as np

# Get argument first
nnPath = str((Path(__file__).parent / Path('../models/mobilenet-ssd_openvino_2021.4_6shave.blob')).resolve().absolute())
if len(sys.argv) > 1:
    nnPath = sys.argv[1]

if not Path(nnPath).exists():
    import sys
    raise FileNotFoundError(f'Required file/s not found, please run "{sys.executable} install_requirements.py"')

# MobilenetSSD label texts
labelMap = ["background", "aeroplane", "bicycle", "bird", "boat", "bottle", "bus", "car", "cat", "chair", "cow",
            "diningtable", "dog", "horse", "motorbike", "person", "pottedplant", "sheep", "sofa", "train", "tvmonitor"]

# Create pipeline
pipeline = dai.Pipeline()

# Define sources and outputs
camRgb = pipeline.create(dai.node.ColorCamera)
videoEncoder = pipeline.create(dai.node.VideoEncoder)
monoRight = pipeline.create(dai.node.MonoCamera)
monoLeft = pipeline.create(dai.node.MonoCamera)
depth = pipeline.create(dai.node.StereoDepth)
manip = pipeline.create(dai.node.ImageManip)
nn = pipeline.create(dai.node.MobileNetDetectionNetwork)

videoOut = pipeline.create(dai.node.XLinkOut)
xoutRight = pipeline.create(dai.node.XLinkOut)
disparityOut = pipeline.create(dai.node.XLinkOut)
manipOut = pipeline.create(dai.node.XLinkOut)
nnOut = pipeline.create(dai.node.XLinkOut)

videoOut.setStreamName('h265')
xoutRight.setStreamName('right')
disparityOut.setStreamName('disparity')
manipOut.setStreamName('manip')
nnOut.setStreamName('nn')

# Properties
camRgb.setBoardSocket(dai.CameraBoardSocket.CAM_A)
camRgb.setResolution(dai.ColorCameraProperties.SensorResolution.THE_1080_P)
monoRight.setCamera("right")
monoRight.setResolution(dai.MonoCameraProperties.SensorResolution.THE_400_P)
monoLeft.setCamera("left")
monoLeft.setResolution(dai.MonoCameraProperties.SensorResolution.THE_400_P)
videoEncoder.setDefaultProfilePreset(30, dai.VideoEncoderProperties.Profile.H265_MAIN)

depth.setDefaultProfilePreset(dai.node.StereoDepth.PresetMode.HIGH_DENSITY)
depth.setRectifyEdgeFillColor(0) # Black, to better see the cutout

nn.setConfidenceThreshold(0.5)
nn.setBlobPath(nnPath)
nn.setNumInferenceThreads(2)
nn.input.setBlocking(False)

# The NN model expects BGR input. By default ImageManip output type would be same as input (gray in this case)
manip.initialConfig.setFrameType(dai.ImgFrame.Type.BGR888p)
manip.initialConfig.setResize(300, 300)

# Linking
camRgb.video.link(videoEncoder.input)
videoEncoder.bitstream.link(videoOut.input)
monoRight.out.link(xoutRight.input)
monoRight.out.link(depth.right)
monoLeft.out.link(depth.left)
depth.disparity.link(disparityOut.input)
depth.rectifiedRight.link(manip.inputImage)
manip.out.link(nn.input)
manip.out.link(manipOut.input)
nn.out.link(nnOut.input)

# Disparity range is used for normalization
disparityMultiplier = 255 / depth.initialConfig.getMaxDisparity()

# Connect to device and start pipeline
with dai.Device(pipeline) as device:

    queueSize = 8
    qRight = device.getOutputQueue("right", queueSize)
    qDisparity = device.getOutputQueue("disparity", queueSize)
    qManip = device.getOutputQueue("manip", queueSize)
    qDet = device.getOutputQueue("nn", queueSize)
    qRgbEnc = device.getOutputQueue('h265', maxSize=30, blocking=True)

    frame = None
    frameManip = None
    frameDisparity = None
    detections = []
    offsetX = (monoRight.getResolutionWidth() - monoRight.getResolutionHeight()) // 2
    color = (255, 0, 0)
    croppedFrame = np.zeros((monoRight.getResolutionHeight(), monoRight.getResolutionHeight()))

    def frameNorm(frame, bbox):
        normVals = np.full(len(bbox), frame.shape[0])
        normVals[::2] = frame.shape[1]
        return (np.clip(np.array(bbox), 0, 1) * normVals).astype(int)

    videoFile = open('video.h265', 'wb')
    cv2.namedWindow("right", cv2.WINDOW_NORMAL)
    cv2.namedWindow("manip", cv2.WINDOW_NORMAL)

    while True:
        inRight = qRight.tryGet()
        inManip = qManip.tryGet()
        inDet = qDet.tryGet()
        inDisparity = qDisparity.tryGet()

        while qRgbEnc.has():
            qRgbEnc.get().getData().tofile(videoFile)

        if inRight is not None:
            frame = inRight.getCvFrame()

        if inManip is not None:
            frameManip = inManip.getCvFrame()

        if inDisparity is not None:
            # Apply color map for better visualization
            frameDisparity = inDisparity.getCvFrame()
            frameDisparity = (frameDisparity*disparityMultiplier).astype(np.uint8)
            frameDisparity = cv2.applyColorMap(frameDisparity, cv2.COLORMAP_JET)

        if inDet is not None:
            detections = inDet.detections

        if frame is not None:
            for detection in detections:
                bbox = frameNorm(croppedFrame, (detection.xmin, detection.ymin, detection.xmax, detection.ymax))
                bbox[::2] += offsetX
                cv2.putText(frame, labelMap[detection.label], (bbox[0] + 10, bbox[1] + 20), cv2.FONT_HERSHEY_TRIPLEX, 0.5, color)
                cv2.putText(frame, f"{int(detection.confidence * 100)}%", (bbox[0] + 10, bbox[1] + 40), cv2.FONT_HERSHEY_TRIPLEX, 0.5, color)
                cv2.rectangle(frame, (bbox[0], bbox[1]), (bbox[2], bbox[3]), color, 2)
            # Show the right cam frame
            cv2.imshow("right", frame)

        if frameDisparity is not None:
            for detection in detections:
                bbox = frameNorm(croppedFrame, (detection.xmin, detection.ymin, detection.xmax, detection.ymax))
                bbox[::2] += offsetX
                cv2.rectangle(frameDisparity, (bbox[0], bbox[1]), (bbox[2], bbox[3]), color, 2)
                cv2.putText(frameDisparity, labelMap[detection.label], (bbox[0] + 10, bbox[1] + 20), cv2.FONT_HERSHEY_TRIPLEX, 0.5, color)
                cv2.putText(frameDisparity, f"{int(detection.confidence * 100)}%", (bbox[0] + 10, bbox[1] + 40), cv2.FONT_HERSHEY_TRIPLEX, 0.5, color)
            # Show the disparity frame
            cv2.imshow("disparity", frameDisparity)

        if frameManip is not None:
            for detection in detections:
                bbox = frameNorm(frameManip, (detection.xmin, detection.ymin, detection.xmax, detection.ymax))
                cv2.rectangle(frameManip, (bbox[0], bbox[1]), (bbox[2], bbox[3]), color, 2)
                cv2.putText(frameManip, labelMap[detection.label], (bbox[0] + 10, bbox[1] + 20), cv2.FONT_HERSHEY_TRIPLEX, 0.5, color)
                cv2.putText(frameManip, f"{int(detection.confidence * 100)}%", (bbox[0] + 10, bbox[1] + 40), cv2.FONT_HERSHEY_TRIPLEX, 0.5, color)
            # Show the manip frame
            cv2.imshow("manip", frameManip)

        if cv2.waitKey(1) == ord('q'):
            break

    print("To view the encoded data, convert the stream file (.h265) into a video file (.mp4) using a command below:")
    print("ffmpeg -framerate 30 -i video.h265 -c copy video.mp4")
```

#### C++

```cpp
#include <cstdio>
#include <iostream>

// Includes common necessary includes for development using depthai library
#include "depthai/depthai.hpp"

// MobilenetSSD label texts
static const std::vector<std::string> labelMap = {"background", "aeroplane", "bicycle",     "bird",  "boat",        "bottle", "bus",
                                                  "car",        "cat",       "chair",       "cow",   "diningtable", "dog",    "horse",
                                                  "motorbike",  "person",    "pottedplant", "sheep", "sofa",        "train",  "tvmonitor"};

int main(int argc, char** argv) {
    using namespace std;
    // Default blob path provided by Hunter private data download
    // Applicable for easier example usage only
    std::string nnPath(BLOB_PATH);

    // If path to blob specified, use that
    if(argc > 1) {
        nnPath = std::string(argv[1]);
    }

    // Print which blob we are using
    printf("Using blob at path: %s\n", nnPath.c_str());

    // Create pipeline
    dai::Pipeline pipeline;

    // Define sources and outputs
    auto camRgb = pipeline.create<dai::node::ColorCamera>();
    auto videoEncoder = pipeline.create<dai::node::VideoEncoder>();
    auto monoRight = pipeline.create<dai::node::MonoCamera>();
    auto monoLeft = pipeline.create<dai::node::MonoCamera>();
    auto depth = pipeline.create<dai::node::StereoDepth>();
    auto manip = pipeline.create<dai::node::ImageManip>();
    auto nn = pipeline.create<dai::node::MobileNetDetectionNetwork>();

    auto videoOut = pipeline.create<dai::node::XLinkOut>();
    auto xoutRight = pipeline.create<dai::node::XLinkOut>();
    auto disparityOut = pipeline.create<dai::node::XLinkOut>();
    auto manipOut = pipeline.create<dai::node::XLinkOut>();
    auto nnOut = pipeline.create<dai::node::XLinkOut>();

    videoOut->setStreamName("h265");
    xoutRight->setStreamName("right");
    disparityOut->setStreamName("disparity");
    manipOut->setStreamName("manip");
    nnOut->setStreamName("nn");

    // Properties
    camRgb->setBoardSocket(dai::CameraBoardSocket::CAM_A);
    camRgb->setResolution(dai::ColorCameraProperties::SensorResolution::THE_1080_P);
    monoRight->setCamera("right");
    monoRight->setResolution(dai::MonoCameraProperties::SensorResolution::THE_400_P);
    monoLeft->setCamera("left");
    monoLeft->setResolution(dai::MonoCameraProperties::SensorResolution::THE_400_P);
    videoEncoder->setDefaultProfilePreset(30, dai::VideoEncoderProperties::Profile::H265_MAIN);

    depth->setDefaultProfilePreset(dai::node::StereoDepth::PresetMode::HIGH_DENSITY);
    depth->setRectifyEdgeFillColor(0);  // Black, to better see the cutout

    nn->setConfidenceThreshold(0.5);
    nn->setBlobPath(nnPath);
    nn->setNumInferenceThreads(2);
    nn->input.setBlocking(false);

    // The NN model expects BGR input-> By default ImageManip output type would be same as input (gray in this case)
    manip->initialConfig.setFrameType(dai::ImgFrame::Type::BGR888p);
    manip->initialConfig.setResize(300, 300);

    // Linking
    camRgb->video.link(videoEncoder->input);
    videoEncoder->bitstream.link(videoOut->input);
    monoRight->out.link(xoutRight->input);
    monoRight->out.link(depth->right);
    monoLeft->out.link(depth->left);
    depth->disparity.link(disparityOut->input);
    depth->rectifiedRight.link(manip->inputImage);
    manip->out.link(nn->input);
    manip->out.link(manipOut->input);
    nn->out.link(nnOut->input);

    // Disparity range is used for normalization
    float disparityMultiplier = 255 / depth->initialConfig.getMaxDisparity();

    // Connect to device and start pipeline
    dai::Device device(pipeline);

    // Queues
    int queueSize = 8;
    auto qRight = device.getOutputQueue("right", queueSize);
    auto qDisparity = device.getOutputQueue("disparity", queueSize);
    auto qManip = device.getOutputQueue("manip", queueSize);
    auto qDet = device.getOutputQueue("nn", queueSize);
    auto qRgbEnc = device.getOutputQueue("h265", 30, true);

    cv::Mat frame;
    cv::Mat frameManip;
    cv::Mat frameDisparity;
    std::vector<dai::ImgDetection> detections;
    int offsetX = (monoRight->getResolutionWidth() - monoRight->getResolutionHeight()) / 2;
    auto color = cv::Scalar(255, 0, 0);

    auto videoFile = std::ofstream("video.h265", std::ios::binary);
    cv::namedWindow("right", cv::WINDOW_NORMAL);
    cv::namedWindow("manip", cv::WINDOW_NORMAL);

    while(true) {
        auto inRight = qRight->tryGet<dai::ImgFrame>();
        auto inManip = qManip->tryGet<dai::ImgFrame>();
        auto inDet = qDet->tryGet<dai::ImgDetections>();
        auto inDisparity = qDisparity->tryGet<dai::ImgFrame>();

        auto out1 = qRgbEnc->get<dai::ImgFrame>();
        videoFile.write((char*)out1->getData().data(), out1->getData().size());

        if(inRight) {
            frame = inRight->getCvFrame();
        }

        if(inManip) {
            frameManip = inManip->getCvFrame();
        }

        if(inDisparity) {
            frameDisparity = inDisparity->getCvFrame();
            frameDisparity.convertTo(frameDisparity, CV_8UC1, disparityMultiplier);
            cv::applyColorMap(frameDisparity, frameDisparity, cv::COLORMAP_JET);
        }

        if(inDet) {
            detections = inDet->detections;
        }

        if(!frame.empty()) {
            for(auto& detection : detections) {
                int x1 = detection.xmin * monoRight->getResolutionHeight() + offsetX;
                int y1 = detection.ymin * monoRight->getResolutionHeight();
                int x2 = detection.xmax * monoRight->getResolutionHeight() + offsetX;
                int y2 = detection.ymax * monoRight->getResolutionHeight();

                uint32_t labelIndex = detection.label;
                std::string labelStr = to_string(labelIndex);
                if(labelIndex < labelMap.size()) {
                    labelStr = labelMap[labelIndex];
                }
                cv::putText(frame, labelStr, cv::Point(x1 + 10, y1 + 20), cv::FONT_HERSHEY_TRIPLEX, 0.5, color);
                std::stringstream confStr;
                confStr << std::fixed << std::setprecision(2) << detection.confidence * 100;
                cv::putText(frame, confStr.str(), cv::Point(x1 + 10, y1 + 40), cv::FONT_HERSHEY_TRIPLEX, 0.5, color);
                cv::rectangle(frame, cv::Rect(cv::Point(x1, y1), cv::Point(x2, y2)), color, cv::FONT_HERSHEY_SIMPLEX);
            }
            // Show the right cam frame
            cv::imshow("right", frame);
        }

        if(!frameDisparity.empty()) {
            for(auto& detection : detections) {
                int x1 = detection.xmin * monoRight->getResolutionHeight() + offsetX;
                int y1 = detection.ymin * monoRight->getResolutionHeight();
                int x2 = detection.xmax * monoRight->getResolutionHeight() + offsetX;
                int y2 = detection.ymax * monoRight->getResolutionHeight();

                uint32_t labelIndex = detection.label;
                std::string labelStr = to_string(labelIndex);
                if(labelIndex < labelMap.size()) {
                    labelStr = labelMap[labelIndex];
                }
                cv::putText(frameDisparity, labelStr, cv::Point(x1 + 10, y1 + 20), cv::FONT_HERSHEY_TRIPLEX, 0.5, color);
                std::stringstream confStr;
                confStr << std::fixed << std::setprecision(2) << detection.confidence * 100;
                cv::putText(frameDisparity, confStr.str(), cv::Point(x1 + 10, y1 + 40), cv::FONT_HERSHEY_TRIPLEX, 0.5, color);
                cv::rectangle(frameDisparity, cv::Rect(cv::Point(x1, y1), cv::Point(x2, y2)), color, cv::FONT_HERSHEY_SIMPLEX);
            }
            // Show the disparity frame
            cv::imshow("disparity", frameDisparity);
        }

        if(!frameManip.empty()) {
            for(auto& detection : detections) {
                int x1 = detection.xmin * frameManip.cols;
                int y1 = detection.ymin * frameManip.rows;
                int x2 = detection.xmax * frameManip.cols;
                int y2 = detection.ymax * frameManip.rows;

                uint32_t labelIndex = detection.label;
                std::string labelStr = to_string(labelIndex);
                if(labelIndex < labelMap.size()) {
                    labelStr = labelMap[labelIndex];
                }
                cv::putText(frameManip, labelStr, cv::Point(x1 + 10, y1 + 20), cv::FONT_HERSHEY_TRIPLEX, 0.5, color);
                std::stringstream confStr;
                confStr << std::fixed << std::setprecision(2) << detection.confidence * 100;
                cv::putText(frameManip, confStr.str(), cv::Point(x1 + 10, y1 + 40), cv::FONT_HERSHEY_TRIPLEX, 0.5, color);
                cv::rectangle(frameManip, cv::Rect(cv::Point(x1, y1), cv::Point(x2, y2)), color, cv::FONT_HERSHEY_SIMPLEX);
            }
            // Show the manip frame
            cv::imshow("manip", frameManip);
        }

        int key = cv::waitKey(1);
        if(key == 'q' || key == 'Q') {
            break;
        }
    }
    cout << "To view the encoded data, convert the stream file (.h265) into a video file (.mp4), using a command below:" << endl;
    cout << "ffmpeg -framerate 30 -i video.h265 -c copy video.mp4" << endl;
    return 0;
}
```

## Pipeline

### Need assistance?

Head over to [Discussion Forum](https://discuss.luxonis.com/) for technical support or any other questions you might have.