doleron

std::vector<cv::Mat> predictions;
net.forward(predictions, net.getUnconnectedOutLayersNames());
const cv::Mat &output = predictions[0];

doleron

def unwrap_detection(input_image, output_data):
    class_ids = []
    confidences = []
    boxes = []

    rows = output_data.shape[0]

    image_width, image_height, _ = input_image.shape

    x_factor = image_width / 640

doleron

struct Detection
{
    int class_id;
    float confidence;
    cv::Rect box;
};

void detect(const cv::Mat &input_image, constcv::Mat &output, std::vector<Detection> &output) {
   
    float x_factor = input_image.cols / 640.;

doleron

indexes = cv2.dnn.NMSBoxes(boxes, confidences, 0.25, 0.45) 

result_class_ids = []
result_confidences = []
result_boxes = []

for i in indexes:
    result_confidences.append(confidences[i])
    result_class_ids.append(class_ids[i])
    result_boxes.append(boxes[i])

doleron

class_list = []
with open("classes.txt", "r") as f:
    class_list = [cname.strip() for cname in f.readlines()]

colors = [(255, 255, 0), (0, 255, 0), (0, 255, 255), (255, 0, 0)]

for i in range(len(result_class_ids)):

    box = result_boxes[i]
    class_id = result_class_ids[i]

doleron

# load the model as usual
net = cv2.dnn.readNet('yolov5s.onnx')

net.setPreferableBackend(cv2.dnn.DNN_BACKEND_CUDA)
net.setPreferableTarget(cv2.dnn.DNN_TARGET_CUDA_FP16)

doleron

auto net = cv::dnn::readNet("yolov5s.onnx");

net.setPreferableBackend(cv::dnn::DNN_BACKEND_CUDA);
net.setPreferableTarget(cv::dnn::DNN_TARGET_CUDA_FP16);

doleron

import cv2

image_path = 'data/full/with_mask (4)'

image = cv2.imread(image_path + '.jpg')

class_list = ['using mask', 'without mask']
colors = [(0, 255, 0), (0, 255, 255)]

height, width, _ = image.shape

doleron

import os, shutil, random

# preparing the folder structure

full_data_path = 'data/obj/'
extension_allowed = '.jpg'
split_percentage = 90

images_path = 'data/images/'
if os.path.exists(images_path):

doleron

const N = 91;
// generating a sequence 91 elements from 0 to 2π
const xs = [...Array(N).keys()].map(x => 2*Math.PI*x / (N - 1));
// generating f(x) = sin(x)
const signal = xs.map(x => Math.sin(x));

// sampling 20 random elements and adding a N(0, 0.4) gaussian error
const sampleSize = 20;
const sample = [...Array(N).keys()].sort(() => 0.5 - jStat.uniform.sample(0, 1)).slice(0, sampleSize).map(index => ({x: index, y: signal[index] + jStat.normal.sample(0.0, 0.4)}));