Невозможно получить вывод обнаружения YOLO на кадрах из камеры - Python Flask - Ошибка - ValueError: невозможно преобразовать число с плавающей запятой NaN в целое число

Question

Я столкнулся с двумя проблемами из-за того, что обнаружение YOLO не отображается в кадре -

1. Ошибка - ValueError: невозможно преобразовать float NaN в целое число

2. Ошибка - w = int (обнаружение [2] * ширина) OverflowError: невозможно преобразовать бесконечность с плавающей запятой в целое число

My Цель состоит в том, чтобы получить доступ к веб-камере на машине клиента, обработать кадр и обнаружить объект в соответствии с предварительно обученной моделью Yolo.

Точная ошибка на терминале Ma c - center_x = int (обнаружение [0] * ширина) ValueError: невозможно преобразовать число с плавающей запятой NaN в целое число

Совместное использование кода для используемых важных файлов - app. запустить, обнаружить. html, detect_processing.py, training_model.py

 run.py
@socketio.on('image')
def image(data_image):

    # decode and convert into image
    b = io.BytesIO(base64.b64decode(data_image))
    pimg = Image.open(b)
    pimg = numpy.array(pimg)
    #print(pimg)

    # Process the image frame
    #frame = imutils.resize(pimg, width=500, height=375)
    frameb = cv2.cvtColor(pimg, cv2.COLOR_RGB2BGR)
    frameb = cv2.flip(frameb, 1)
    frameb = detect_object(frameb)
    imgencode = cv2.imencode('.jpg', frameb)[1]

    # base64 encode
    stringData = base64.b64encode(imgencode).decode('utf-8')
    b64_src = 'data:image/jpg;base64,'
    stringData = b64_src + stringData

    # emit the frame back
    emit('response_back', stringData)


# main driver function  
if __name__ == '__main__': 
    socketio.run(app,debug=True)

---

---

detect_processing.py
import cv2
import numpy as np
from trained_model import *


def detect_object(frame):

        # DO WHAT YOU WANT WITH TENSORFLOW / KERAS AND OPENCV

        # Implement YOLO object detection  model
        # keep outside while loop so that we define it once and not repeat defining it in the loop
        font=cv2.FONT_HERSHEY_SIMPLEX
        height, width, channels = frame.shape
        print(width, height)


        #Detecing images using cv2 neural network (We can change the processing size of input frame for improving speed on cpu)
        blob = cv2.dnn.blobFromImage(frame,0.00392,(320,320),(0,0,0),True,crop = False)

        #passing blob through the network to detect and prediction
        #Outs is an array that conains all the informations about objects detected,
        #their position and the confidence about the detection.
        net.setInput(blob)
        outs = net.forward(output_layer)



        # Showing informations on the screen
        class_ids = []
        confidences = []
        boxes = []
        for out in outs:
            for detection in out:
                    scores = detection[5:]
                    class_id = np.argmax(scores)
                    confidence = scores[class_id]
                    if confidence > 0.5:
                        # Object detected
                        center_x = int(detection[0] * width)
                        center_y = int(detection[1] * height)
                        w = int(detection[2] * width)
                        h = int(detection[3] * height)

                        # Rectangle coordinates
                        x = int(center_x - w / 2)
                        y = int(center_y - h / 2)

                        boxes.append([x, y, w, h])
                        confidences.append(float(confidence))
                        class_ids.append(class_id)
                        text = "{:.2f}%".format(confidence * 100)


        #When we perform the detection, it happens that we have more boxes for the same object, so we should use another function to remove this “noise”.
        #It’s called Non maximum suppresion.
        indexes = cv2.dnn.NMSBoxes(boxes, confidences, 0.5, 0.4)



        #Box: contain the coordinates of the rectangle sorrounding the object detected.
        #Label: it’s the name of the object detected
        #Confidence: the confidence about the detection from 0 to 1

        for i in range(len(boxes)):
            if i in indexes:
                x, y, w, h = boxes[i]
                label = str(classes[class_ids[i]])
                label_plus = label + " - " + text
                #label_plus = label_plus[::-1]
                #frame=cv2.flip(frame,1)
                cv2.rectangle(frame, (x, y), (x + w, y + h), (0,0,255), 3)
                cv2.putText(frame, label_plus, (x, y - 20), font, 0.6, (0,255,255), 1)


        return frame

---

 trained_model.py
import cv2

#we have choosed the yolo version which works best on image of size 416*416
#we can use yolov3 - tiny version as it has been optimized for CPU but detection won't be that great
path1 = "/Users/sonijha/Downloads/Yolo_ObjectDetection/yolov3_416.weights"
path2 = "/Users/sonijha/Downloads/Yolo_ObjectDetection/yolov3.cfg"
net = cv2.dnn.readNet(path1, path2)

#Reading from coco.names for 80 objects which are pre-trained by yolo
#with statement in Python is used in exception handling to make the code cleaner and much more readable.
classes = []
with open ("coco.names", "r") as f:
        classes = [line.strip() for line in f.readlines()]

print(classes[73])

layer_names = net.getLayerNames()

#getLayerNames(): Get the name of all layers of the network.
#getUnconnectedOutLayers(): Get the index of the output layers.

output_layer = [layer_names[i[0]-1] for i in net.getUnconnectedOutLayers()]