objectdetection.py

import cv2
import numpy as np



image = cv2.imread("./pic.jpg")


Width = image.shape[1]
Height = image.shape[0]
scale = 0.00392


classes = None
with open("./config/coco.names", 'r') as f:
    classes = [line.strip() for line in f.readlines()]

COLORS = np.random.uniform(0, 255, size=(len(classes), 3))



def get_output_layers(net):
    layer_names = net.getLayerNames()
    print((layer_names))
    print(layer_names[199])
    print(net.getUnconnectedOutLayers())
    # output_layers = [layer_names[i[0] - 1] for i in net.getUnconnectedOutLayers()]
    output_layers = [layer_names[200-1],layer_names[227-1],layer_names[254-1]]
    return output_layers

def draw_bounding_box(img, class_id, confidence, x, y, x_plus_w, y_plus_h):
    label = str(classes[class_id])
    color = COLORS[class_id]
    cv2.rectangle(img, (x,y), (x_plus_w,y_plus_h), color, 2)
    cv2.putText(img, label, (x-10,y-10), cv2.FONT_HERSHEY_SIMPLEX, 0.5, color, 2)

net = cv2.dnn.readNet("./config/yolov3.weights", "./config/yolov3.cfg")
blob = cv2.dnn.blobFromImage(image, scale, (416,416), (0,0,0), True, crop=False)

net.setInput(blob)



outs = net.forward(get_output_layers(net))

# initialization
class_ids = []
confidences = []
boxes = []
conf_threshold = 0.5
nms_threshold = 0.4


print(outs)
for out in outs:
    for detection in out:
        try:
            scores = detection[5:]
        except:
            scores = detection
        
        class_id = np.argmax(scores)
        try:
            confidence = scores[class_id]
        except:
            confidence = scores
        if confidence > 0.5:
            print(f"{confidence}nnnnn")
            center_x = int(detection[0] * Width)
            center_y = int(detection[1] * Height)
            w = int(detection[2] * Width)
            h = int(detection[3] * Height)
            x = center_x - w / 2
            y = center_y - h / 2
            class_ids.append(class_id)
            confidences.append(float(confidence))
            boxes.append([x, y, w, h])


indices = cv2.dnn.NMSBoxes(boxes, confidences, conf_threshold, nms_threshold)



print(indices)
for i in indices:
    # i = i[0]
    box = boxes[i]
    x = box[0]
    y = box[1]
    w = box[2]
    h = box[3]
    
    draw_bounding_box(image, class_ids[i], confidences[i], round(x), round(y), round(x+w), round(y+h))

   
cv2.imshow("object detection", image)
cv2.waitKey()
cv2.imwrite("object-detection.jpg", image)
cv2.destroyAllWindows()