RGB & TinyYoloV3 decoding on device

This example shows how to run TinyYoloV3 on the RGB input frame, and how to display both the RGB preview and the metadata results from the TinyYoloV3 on the preview. Decoding is done on Myriad instead on the host.

Configurable, network dependent parameters are required for correct decoding: setNumClasses - number of YOLO classes setCoordinateSize - size of coordinate setAnchors - yolo anchors setAnchorMasks - anchorMasks26, anchorMasks13 (anchorMasks52 - additionally for full YOLOv3) 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 YoloV3-tiny blob (tiny-yolo-v3_openvino_2021.2_6shave.blob file) to work - 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
#!/usr/bin/env python3

"""
Tiny-yolo-v3 device side decoding demo
  YOLO v3 Tiny is a real-time object detection model implemented with Keras* from
  this repository <https://github.com/david8862/keras-YOLOv3-model-set> and converted
  to TensorFlow* framework. This model was pretrained on COCO* dataset with 80 classes.
"""

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

# Get argument first
nnPath = str((Path(__file__).parent / Path('models/yolo-v3-tiny-tf_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"')

# Tiny yolo v3 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()
detectionNetwork = pipeline.createYoloDetectionNetwork()
xoutRgb = pipeline.createXLinkOut()
nnOut = pipeline.createXLinkOut()

xoutRgb.setStreamName("rgb")
nnOut.setStreamName("nn")

# 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)

# Network specific settings
detectionNetwork.setConfidenceThreshold(0.5)
detectionNetwork.setNumClasses(80)
detectionNetwork.setCoordinateSize(4)
detectionNetwork.setAnchors(np.array([10, 14, 23, 27, 37, 58, 81, 82, 135, 169, 344, 319]))
detectionNetwork.setAnchorMasks({"side26": np.array([1, 2, 3]), "side13": np.array([3, 4, 5])})
detectionNetwork.setIouThreshold(0.5)
detectionNetwork.setBlobPath(nnPath)
detectionNetwork.setNumInferenceThreads(2)
detectionNetwork.input.setBlocking(False)

# Linking
camRgb.preview.link(detectionNetwork.input)
if syncNN:
    detectionNetwork.passthrough.link(xoutRgb.input)
else:
    camRgb.preview.link(xoutRgb.input)

detectionNetwork.out.link(nnOut.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
    qRgb = device.getOutputQueue(name="rgb", maxSize=4, blocking=False)
    qDet = device.getOutputQueue(name="nn", maxSize=4, blocking=False)

    frame = None
    detections = []
    startTime = time.monotonic()
    counter = 0
    color2 = (255, 255, 255)

    # nn data, being the bounding box locations, are in <0..1> range - they need to be normalized with frame width/height
    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)

    def displayFrame(name, frame):
        color = (255, 0, 0)
        for detection in detections:
            bbox = frameNorm(frame, (detection.xmin, detection.ymin, detection.xmax, detection.ymax))
            cv2.putText(frame, labelMap[detection.label], (bbox[0] + 10, bbox[1] + 20), cv2.FONT_HERSHEY_TRIPLEX, 0.5, 255)
            cv2.putText(frame, f"{int(detection.confidence * 100)}%", (bbox[0] + 10, bbox[1] + 40), cv2.FONT_HERSHEY_TRIPLEX, 0.5, 255)
            cv2.rectangle(frame, (bbox[0], bbox[1]), (bbox[2], bbox[3]), color, 2)
        # Show the frame
        cv2.imshow(name, frame)

    while True:
        if syncNN:
            inRgb = qRgb.get()
            inDet = qDet.get()
        else:
            inRgb = qRgb.tryGet()
            inDet = qDet.tryGet()

        if inRgb is not None:
            frame = inRgb.getCvFrame()
            cv2.putText(frame, "NN fps: {:.2f}".format(counter / (time.monotonic() - startTime)),
                        (2, frame.shape[0] - 4), cv2.FONT_HERSHEY_TRIPLEX, 0.4, color2)

        if inDet is not None:
            detections = inDet.detections
            counter += 1

        if frame is not None:
            displayFrame("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
#include <chrono>
#include <iostream>

// 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 detectionNetwork = pipeline.create<dai::node::YoloDetectionNetwork>();
    auto xoutRgb = pipeline.create<dai::node::XLinkOut>();
    auto nnOut = pipeline.create<dai::node::XLinkOut>();

    xoutRgb->setStreamName("rgb");
    nnOut->setStreamName("nn");

    // 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);

    // Network specific settings
    detectionNetwork->setConfidenceThreshold(0.5f);
    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(0.5f);
    detectionNetwork->setBlobPath(nnPath);
    detectionNetwork->setNumInferenceThreads(2);
    detectionNetwork->input.setBlocking(false);

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

    detectionNetwork->out.link(nnOut->input);

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

    auto qRgb = device.getOutputQueue("rgb", 4, false);
    auto qDet = device.getOutputQueue("nn", 4, false);

    cv::Mat frame;
    std::vector<dai::ImgDetection> detections;
    auto startTime = steady_clock::now();
    int counter = 0;
    float fps = 0;
    auto color2 = cv::Scalar(255, 255, 255);

    // Add bounding boxes and text to the frame and show it to the user
    auto displayFrame = [](std::string name, cv::Mat frame, std::vector<dai::ImgDetection>& detections) {
        auto color = cv::Scalar(255, 0, 0);
        // nn data, being the bounding box locations, are in <0..1> range - they need to be normalized with frame width/height
        for(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, 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 frame
        cv::imshow(name, frame);
    };

    while(true) {
        std::shared_ptr<dai::ImgFrame> inRgb;
        std::shared_ptr<dai::ImgDetections> inDet;

        if(syncNN) {
            inRgb = qRgb->get<dai::ImgFrame>();
            inDet = qDet->get<dai::ImgDetections>();
        } else {
            inRgb = qRgb->tryGet<dai::ImgFrame>();
            inDet = qDet->tryGet<dai::ImgDetections>();
        }

        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;
        }

        if(inRgb) {
            frame = inRgb->getCvFrame();
            std::stringstream fpsStr;
            fpsStr << "NN fps: " << std::fixed << std::setprecision(2) << fps;
            cv::putText(frame, fpsStr.str(), cv::Point(2, inRgb->getHeight() - 4), cv::FONT_HERSHEY_TRIPLEX, 0.4, color2);
        }

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

        if(!frame.empty()) {
            displayFrame("rgb", frame, detections);
        }

        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.