RGB & TinyYolo with spatial data

This example shows how to run TinyYoloV3 and v4 on the RGB input frame, and how to display both the RGB preview, detections, depth map and spatial information (X,Y,Z). It’s similar to example RGB & MobilenetSSD with spatial data except it is running TinyYolo network. X,Y,Z coordinates are relative to the center of depth map.

setNumClasses - number of YOLO classes setCoordinateSize - size of coordinate setAnchors - yolo anchors setAnchorMasks - anchorMasks26, anchorMasks13 (anchorMasks52 - additionally for full YOLOv4) setIouThreshold - intersection over union threshold setConfidenceThreshold - confidence threshold above which objects are detected

Similiar samples:

Demo

Setup

Please run the install script to download all required dependencies. Please note that this script must be ran from git context, so you have to download the depthai-python repository first and then run the script

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

For additional information, please follow installation guide

This example also requires YOLOv4-tiny blob (tiny-yolo-v4_openvino_2021.2_6shave.blob file) to work - you can download it from here

YOLOv3-tiny blob (tiny-yolo-v3_openvino_2021.2_6shave.blob file) can be used too - you can download it from here

Source code

Also available on GitHub

  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
#!/usr/bin/env python3

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

'''
Spatial Tiny-yolo example
  Performs inference on RGB camera and retrieves spatial location coordinates: x,y,z relative to the center of depth map.
  Can be used for tiny-yolo-v3 or tiny-yolo-v4 networks
'''

# Get argument first
nnBlobPath = str((Path(__file__).parent / Path('models/yolo-v4-tiny-tf_openvino_2021.4_6shave.blob')).resolve().absolute())
if len(sys.argv) > 1:
    nnBlobPath = sys.argv[1]

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

# Tiny yolo v3/4 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"
]

syncNN = True

# Create pipeline
pipeline = dai.Pipeline()

# Define sources and outputs
camRgb = pipeline.createColorCamera()
spatialDetectionNetwork = pipeline.createYoloSpatialDetectionNetwork()
monoLeft = pipeline.createMonoCamera()
monoRight = pipeline.createMonoCamera()
stereo = pipeline.createStereoDepth()

xoutRgb = pipeline.createXLinkOut()
xoutNN = pipeline.createXLinkOut()
xoutBoundingBoxDepthMapping = pipeline.createXLinkOut()
xoutDepth = pipeline.createXLinkOut()

xoutRgb.setStreamName("rgb")
xoutNN.setStreamName("detections")
xoutBoundingBoxDepthMapping.setStreamName("boundingBoxDepthMapping")
xoutDepth.setStreamName("depth")

# Properties
camRgb.setPreviewSize(416, 416)
camRgb.setResolution(dai.ColorCameraProperties.SensorResolution.THE_1080_P)
camRgb.setInterleaved(False)
camRgb.setColorOrder(dai.ColorCameraProperties.ColorOrder.BGR)

monoLeft.setResolution(dai.MonoCameraProperties.SensorResolution.THE_400_P)
monoLeft.setBoardSocket(dai.CameraBoardSocket.LEFT)
monoRight.setResolution(dai.MonoCameraProperties.SensorResolution.THE_400_P)
monoRight.setBoardSocket(dai.CameraBoardSocket.RIGHT)

# setting node configs
stereo.initialConfig.setConfidenceThreshold(255)

spatialDetectionNetwork.setBlobPath(nnBlobPath)
spatialDetectionNetwork.setConfidenceThreshold(0.5)
spatialDetectionNetwork.input.setBlocking(False)
spatialDetectionNetwork.setBoundingBoxScaleFactor(0.5)
spatialDetectionNetwork.setDepthLowerThreshold(100)
spatialDetectionNetwork.setDepthUpperThreshold(5000)

# Yolo specific parameters
spatialDetectionNetwork.setNumClasses(80)
spatialDetectionNetwork.setCoordinateSize(4)
spatialDetectionNetwork.setAnchors(np.array([10,14, 23,27, 37,58, 81,82, 135,169, 344,319]))
spatialDetectionNetwork.setAnchorMasks({ "side26": np.array([1,2,3]), "side13": np.array([3,4,5]) })
spatialDetectionNetwork.setIouThreshold(0.5)

# Linking
monoLeft.out.link(stereo.left)
monoRight.out.link(stereo.right)

camRgb.preview.link(spatialDetectionNetwork.input)
if syncNN:
    spatialDetectionNetwork.passthrough.link(xoutRgb.input)
else:
    camRgb.preview.link(xoutRgb.input)

spatialDetectionNetwork.out.link(xoutNN.input)
spatialDetectionNetwork.boundingBoxMapping.link(xoutBoundingBoxDepthMapping.input)

stereo.depth.link(spatialDetectionNetwork.inputDepth)
spatialDetectionNetwork.passthroughDepth.link(xoutDepth.input)

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

    # Output queues will be used to get the rgb frames and nn data from the outputs defined above
    previewQueue = device.getOutputQueue(name="rgb", maxSize=4, blocking=False)
    detectionNNQueue = device.getOutputQueue(name="detections", maxSize=4, blocking=False)
    xoutBoundingBoxDepthMappingQueue = device.getOutputQueue(name="boundingBoxDepthMapping", maxSize=4, blocking=False)
    depthQueue = device.getOutputQueue(name="depth", maxSize=4, blocking=False)

    startTime = time.monotonic()
    counter = 0
    fps = 0
    color = (255, 255, 255)

    while True:
        inPreview = previewQueue.get()
        inDet = detectionNNQueue.get()
        depth = depthQueue.get()

        frame = inPreview.getCvFrame()
        depthFrame = depth.getFrame()
        depthFrameColor = cv2.normalize(depthFrame, None, 255, 0, cv2.NORM_INF, cv2.CV_8UC1)
        depthFrameColor = cv2.equalizeHist(depthFrameColor)
        depthFrameColor = cv2.applyColorMap(depthFrameColor, cv2.COLORMAP_HOT)

        counter+=1
        current_time = time.monotonic()
        if (current_time - startTime) > 1 :
            fps = counter / (current_time - startTime)
            counter = 0
            startTime = current_time

        detections = inDet.detections
        if len(detections) != 0:
            boundingBoxMapping = xoutBoundingBoxDepthMappingQueue.get()
            roiDatas = boundingBoxMapping.getConfigData()

            for roiData in roiDatas:
                roi = roiData.roi
                roi = roi.denormalize(depthFrameColor.shape[1], depthFrameColor.shape[0])
                topLeft = roi.topLeft()
                bottomRight = roi.bottomRight()
                xmin = int(topLeft.x)
                ymin = int(topLeft.y)
                xmax = int(bottomRight.x)
                ymax = int(bottomRight.y)

                cv2.rectangle(depthFrameColor, (xmin, ymin), (xmax, ymax), color, cv2.FONT_HERSHEY_SCRIPT_SIMPLEX)


        # If the frame is available, draw bounding boxes on it and show the frame
        height = frame.shape[0]
        width  = frame.shape[1]
        for detection in detections:
            # Denormalize bounding box
            x1 = int(detection.xmin * width)
            x2 = int(detection.xmax * width)
            y1 = int(detection.ymin * height)
            y2 = int(detection.ymax * height)
            try:
                label = labelMap[detection.label]
            except:
                label = detection.label
            cv2.putText(frame, str(label), (x1 + 10, y1 + 20), cv2.FONT_HERSHEY_TRIPLEX, 0.5, 255)
            cv2.putText(frame, "{:.2f}".format(detection.confidence*100), (x1 + 10, y1 + 35), cv2.FONT_HERSHEY_TRIPLEX, 0.5, 255)
            cv2.putText(frame, f"X: {int(detection.spatialCoordinates.x)} mm", (x1 + 10, y1 + 50), cv2.FONT_HERSHEY_TRIPLEX, 0.5, 255)
            cv2.putText(frame, f"Y: {int(detection.spatialCoordinates.y)} mm", (x1 + 10, y1 + 65), cv2.FONT_HERSHEY_TRIPLEX, 0.5, 255)
            cv2.putText(frame, f"Z: {int(detection.spatialCoordinates.z)} mm", (x1 + 10, y1 + 80), 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("depth", depthFrameColor)
        cv2.imshow("rgb", frame)

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

Also available on GitHub

  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
#include <chrono>

#include "utility.hpp"

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

static const std::vector<std::string> 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"};

static std::atomic<bool> syncNN{true};

int main(int argc, char** argv) {
    using namespace std;
    using namespace std::chrono;
    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 spatialDetectionNetwork = pipeline.create<dai::node::YoloSpatialDetectionNetwork>();
    auto monoLeft = pipeline.create<dai::node::MonoCamera>();
    auto monoRight = pipeline.create<dai::node::MonoCamera>();
    auto stereo = pipeline.create<dai::node::StereoDepth>();

    auto xoutRgb = pipeline.create<dai::node::XLinkOut>();
    auto xoutNN = pipeline.create<dai::node::XLinkOut>();
    auto xoutBoundingBoxDepthMapping = pipeline.create<dai::node::XLinkOut>();
    auto xoutDepth = pipeline.create<dai::node::XLinkOut>();

    xoutRgb->setStreamName("rgb");
    xoutNN->setStreamName("detections");
    xoutBoundingBoxDepthMapping->setStreamName("boundingBoxDepthMapping");
    xoutDepth->setStreamName("depth");

    // Properties
    camRgb->setPreviewSize(416, 416);
    camRgb->setResolution(dai::ColorCameraProperties::SensorResolution::THE_1080_P);
    camRgb->setInterleaved(false);
    camRgb->setColorOrder(dai::ColorCameraProperties::ColorOrder::BGR);

    monoLeft->setResolution(dai::MonoCameraProperties::SensorResolution::THE_400_P);
    monoLeft->setBoardSocket(dai::CameraBoardSocket::LEFT);
    monoRight->setResolution(dai::MonoCameraProperties::SensorResolution::THE_400_P);
    monoRight->setBoardSocket(dai::CameraBoardSocket::RIGHT);

    /// setting node configs
    stereo->initialConfig.setConfidenceThreshold(255);

    spatialDetectionNetwork->setBlobPath(nnPath);
    spatialDetectionNetwork->setConfidenceThreshold(0.5f);
    spatialDetectionNetwork->input.setBlocking(false);
    spatialDetectionNetwork->setBoundingBoxScaleFactor(0.5);
    spatialDetectionNetwork->setDepthLowerThreshold(100);
    spatialDetectionNetwork->setDepthUpperThreshold(5000);

    // yolo specific parameters
    spatialDetectionNetwork->setNumClasses(80);
    spatialDetectionNetwork->setCoordinateSize(4);
    spatialDetectionNetwork->setAnchors({10, 14, 23, 27, 37, 58, 81, 82, 135, 169, 344, 319});
    spatialDetectionNetwork->setAnchorMasks({{"side26", {1, 2, 3}}, {"side13", {3, 4, 5}}});
    spatialDetectionNetwork->setIouThreshold(0.5f);

    // Linking
    monoLeft->out.link(stereo->left);
    monoRight->out.link(stereo->right);

    camRgb->preview.link(spatialDetectionNetwork->input);
    if(syncNN) {
        spatialDetectionNetwork->passthrough.link(xoutRgb->input);
    } else {
        camRgb->preview.link(xoutRgb->input);
    }

    spatialDetectionNetwork->out.link(xoutNN->input);
    spatialDetectionNetwork->boundingBoxMapping.link(xoutBoundingBoxDepthMapping->input);

    stereo->depth.link(spatialDetectionNetwork->inputDepth);
    spatialDetectionNetwork->passthroughDepth.link(xoutDepth->input);

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

    // Output queues will be used to get the rgb frames and nn data from the outputs defined above
    auto previewQueue = device.getOutputQueue("rgb", 4, false);
    auto detectionNNQueue = device.getOutputQueue("detections", 4, false);
    auto xoutBoundingBoxDepthMappingQueue = device.getOutputQueue("boundingBoxDepthMapping", 4, false);
    auto depthQueue = device.getOutputQueue("depth", 4, false);

    auto startTime = steady_clock::now();
    int counter = 0;
    float fps = 0;
    auto color = cv::Scalar(255, 255, 255);

    while(true) {
        auto imgFrame = previewQueue->get<dai::ImgFrame>();
        auto inDet = detectionNNQueue->get<dai::SpatialImgDetections>();
        auto depth = depthQueue->get<dai::ImgFrame>();

        cv::Mat frame = imgFrame->getCvFrame();
        cv::Mat depthFrame = depth->getFrame();
        cv::Mat depthFrameColor;
        cv::normalize(depthFrame, depthFrameColor, 255, 0, cv::NORM_INF, CV_8UC1);
        cv::equalizeHist(depthFrameColor, depthFrameColor);
        cv::applyColorMap(depthFrameColor, depthFrameColor, cv::COLORMAP_HOT);

        counter++;
        auto currentTime = steady_clock::now();
        auto elapsed = duration_cast<duration<float>>(currentTime - startTime);
        if(elapsed > seconds(1)) {
            fps = counter / elapsed.count();
            counter = 0;
            startTime = currentTime;
        }

        auto detections = inDet->detections;
        if(!detections.empty()) {
            auto boundingBoxMapping = xoutBoundingBoxDepthMappingQueue->get<dai::SpatialLocationCalculatorConfig>();
            auto roiDatas = boundingBoxMapping->getConfigData();

            for(auto roiData : roiDatas) {
                auto roi = roiData.roi;
                roi = roi.denormalize(depthFrameColor.cols, depthFrameColor.rows);
                auto topLeft = roi.topLeft();
                auto bottomRight = roi.bottomRight();
                auto xmin = (int)topLeft.x;
                auto ymin = (int)topLeft.y;
                auto xmax = (int)bottomRight.x;
                auto ymax = (int)bottomRight.y;

                cv::rectangle(depthFrameColor, cv::Rect(cv::Point(xmin, ymin), cv::Point(xmax, ymax)), color, cv::FONT_HERSHEY_SIMPLEX);
            }
        }

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

            int 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, 255);
            std::stringstream confStr;
            confStr << std::fixed << std::setprecision(2) << detection.confidence * 100;
            cv::putText(frame, confStr.str(), cv::Point(x1 + 10, y1 + 35), cv::FONT_HERSHEY_TRIPLEX, 0.5, 255);

            std::stringstream depthX;
            depthX << "X: " << (int)detection.spatialCoordinates.x << " mm";
            cv::putText(frame, depthX.str(), cv::Point(x1 + 10, y1 + 50), cv::FONT_HERSHEY_TRIPLEX, 0.5, 255);
            std::stringstream depthY;
            depthY << "Y: " << (int)detection.spatialCoordinates.y << " mm";
            cv::putText(frame, depthY.str(), cv::Point(x1 + 10, y1 + 65), cv::FONT_HERSHEY_TRIPLEX, 0.5, 255);
            std::stringstream depthZ;
            depthZ << "Z: " << (int)detection.spatialCoordinates.z << " mm";
            cv::putText(frame, depthZ.str(), cv::Point(x1 + 10, y1 + 80), cv::FONT_HERSHEY_TRIPLEX, 0.5, 255);

            cv::rectangle(frame, cv::Rect(cv::Point(x1, y1), cv::Point(x2, y2)), color, cv::FONT_HERSHEY_SIMPLEX);
        }

        std::stringstream fpsStr;
        fpsStr << std::fixed << std::setprecision(2) << fps;
        cv::putText(frame, fpsStr.str(), cv::Point(2, imgFrame->getHeight() - 4), cv::FONT_HERSHEY_TRIPLEX, 0.4, color);

        cv::imshow("depth", depthFrameColor);
        cv::imshow("rgb", frame);

        int key = cv::waitKey(1);
        if(key == 'q' || key == 'Q') {
            return 0;
        }
    }
    return 0;
}

Got questions?

We’re always happy to help with code or other questions you might have.