if (hasattr(colors_classes, "colors") and

len(colors_classes.colors) == num_classes):

return colors_classes.colors

hsv_tuples = [(x / num_classes, 1., 1.) for x in range(num_classes)]

colors = list(map(lambda x: colorsys.hsv_to_rgb(*x), hsv_tuples))

colors = list(

map(lambda x: (int(x[0] * 255), int(x[1] * 255), int(x[2] * 255)),

colors))

random.seed(10101) # Fixed seed for consistent colors across runs.

random.shuffle(colors) # Shuffle colors to decorrelate adjacent classes.

random.seed(None) # Reset seed to default.

colors_classes.colors = colors # Save colors for future calls.

ymin1, xmin1, ymax1, xmax1 = box1

ymin2, xmin2, ymax2, xmax2 = box2

# 获取矩形框交集对应的左上角和右下角的坐标（intersection）

xx1 = np.max([xmin1, xmin2])

yy1 = np.max([ymin1, ymin2])

xx2 = np.min([xmax1, xmax2])

yy2 = np.min([ymax1, ymax2])

# 计算两个矩形框面积

area1 = (xmax1 - xmin1) * (ymax1 - ymin1)

area2 = (xmax2 - xmin2) * (ymax2 - ymin2)

inter_area = (np.max([0, xx2 - xx1])) * (np.max([0, yy2 - yy1]))

# 计算交集面积

iou = inter_area / (area1 + area2 - inter_area + 1e-6)

# 计算交并比

perhat = np.zeros(shape=(numPeople, numHat))

for perindex in range(numPeople):

for hatindex in range(numHat):

perhat[perindex][hatindex] = compute_iou(boxFilterPerson[perindex], boxFilterHat[hatindex])

#def __init__(self,newname)：self.name=newname,通过访问self.name的形式给实例中增加了name变量，并给name赋了初值newname

def __init__(self):

self.last_rstate = 0

self.last_cov = 0

self.frames = 0

self.trace_id = 0

xAxisDis = np.zeros(shape=(len(centersPerson), len(centersHat)))

for perindex in range(len(centersPerson)):

for hatindex in range(len(centersHat)):

xAxisDis[perindex][hatindex] = centersPerson[perindex][0][0] - centersHat[hatindex][0][0]

# print(np.fabs(xAxisDis).min(1))

print(out_boxes, out_classes, out_scores)

dict = {}

for key in out_classes:

dict[key] = dict.get(key, 0) + 1

print(len(dict))

if len(dict) == 0:

return 000, 000

if len(dict) == 1:

for PersonIndex in range(len(out_boxes)):

RecDraw.append(out_boxes[PersonIndex])

return 666, 666

# elif dict[0] > dict[1]:

# return 666, 666 # 简单版本的就是人比帽子多直接警告，复杂一点就直接pass交给后面的任务

boxFilterPerson = []

boxFilterHat = []

centersPerson = []

centersHat = []

for div_box, div_class, div_score in zip(out_boxes, out_classes, out_scores):

if (div_score >= score_thres) and (div_class == 0):

boxFilterPerson.append(div_box)

centre = np.array([[(div_box[1] + div_box[3]) // 2], [(div_box[0] + div_box[2]) // 2]])

centersPerson.append(centre)

if (div_score >= score_thres) and (div_class == 1):

boxFilterHat.append(div_box)

centre_hat = np.array([[(div_box[1] + div_box[3]) // 2], [(div_box[0] + div_box[2]) // 2]])

centersHat.append(centre_hat)

numPeople = len(centersPerson)

numHat = len(centersHat)

perhat = doiou(boxFilterPerson, boxFilterHat, numPeople, numHat)

m = perhat.sum(axis=1)

for index in range(len(m)):

if m[index] == 0:

RecDraw.append(boxFilterPerson[index])

return 666, 666

xAxisDis = calDistance(centersPerson, centersHat)

dis = np.fabs(xAxisDis).min(1)

for dis_index in range(len(dis)):

if m[dis_index] > 20:

RecDraw.append(boxFilterPerson[index])

return 666, 666

def __init__(self):

self.count = 0

self.dotracking = []

self.str_location = []

self.distance_max = 200

# """注意矩阵的行列此处并没有进行修改"""

def tracking(self, centers, num, image, mode): # 先尝试着单目标的追踪

font = cv2.FONT_HERSHEY_SIMPLEX

if mode == 0:

if centers == 000 and num == 000:

cv2.putText(image, "need check", (11, 33), font, 1, [230, 0, 0], 2)

if centers == 666 and num == 666:

cv2.putText(image, "detecting", (11, 11 + 22), font, 1, [230, 0, 0], 2)

for index in range(len(RecDraw)):

cv2.rectangle(image, (RecDraw[index][1], RecDraw[index][0]), (RecDraw[index][3], RecDraw[index][2]), (230, 0, 0), 2)

elif centers != 666 or num != 666:

cv2.putText(image, "need check", (11, 33), font, 1, [230, 0, 0], 2)

elif mode == 1:

if centers == 000 and num == 000:

cv2.putText(image, "need check", (11, 33), font, 1, [0, 0, 230], 2)

if centers == 666 and num == 666:

cv2.putText(image, "detecting", (11, 11 + 22), font, 1, [0, 0, 230], 2)

for index in range(len(RecDraw)):

cv2.rectangle(image, (RecDraw[index][1], RecDraw[index][0]), (RecDraw[index][3], RecDraw[index][2]),

(0, 0, 230), 2)

elif centers != 666 or num != 666:

"model_path": 'logs/000/trained_weights_final.h5',

"anchors_path": 'model_data/yolo_anchors.txt',

"classes_path": 'model_data/coco_classes.txt',

"score": 0.50,

"iou": 0.3,

"model_image_size": (416, 416),

"gpu_num": 1,

mode = 0

RecDraw.clear()

if pic_path == 0:

yolov3 = YOLO(**yolov3_args)

for i in range(len(ImageDir)):

RecDraw.clear()

ImagePath = "D:/test/testimages/" + ImageDir[i]

ImageName = "D:/test/image/" + str(i) + ".jpg"

img = Image.open(ImagePath)

image, boxes, scores, classes = yolov3.detect_image_mul(img)

centers, num = cal_centre(boxes, classes, scores, 0.05)

result = np.asarray(image)

tracker = dokalman()

tracker.tracking(centers, num, result, mode)

image_bgr = cv2.cvtColor(np.asarray(result), cv2.COLOR_RGB2BGR)

cv2.imwrite(ImageName, image_bgr)

elif pic_path != 0:

yolov3 = YOLO(**yolov3_args)

img = Image.open(pic_path)

image, boxes, scores, classes = yolov3.detect_image_mul(img)

centers, num = cal_centre(boxes, classes, scores, 0.05)

result = np.asarray(image)

tracker = dokalman()

tracker.tracking(centers, num, result, mode)

# print("look there!", RecDraw)

image_bgr = cv2.cvtColor(np.asarray(result), cv2.COLOR_RGB2BGR)

cv2.imwrite("D:/test/pp30.jpg", image_bgr)

cv2.imshow("re", image_bgr)

cv2.waitKey(0)

mode = 1

yolov3 = YOLO(**yolov3_args)

video_cap = cv2.VideoCapture(video_path)

if not video_cap.isOpened():

raise IOError

video_FourCC = int(video_cap.get(cv2.CAP_PROP_FOURCC))

video_fps = video_cap.get(cv2.CAP_PROP_FPS)

video_size = (int(video_cap.get(cv2.CAP_PROP_FRAME_WIDTH)),

int(video_cap.get(cv2.CAP_PROP_FRAME_HEIGHT)))

# video_size = (int(video_cap.get(cv2.CAP_PROP_FRAME_HEIGHT)), int(video_cap.get(cv2.CAP_PROP_FRAME_WIDTH)))

isOutput = True if output_path != "" else False

if isOutput:

out = cv2.VideoWriter(output_path, video_FourCC, video_fps, video_size)

tracker = dokalman()

frame_index = 0

name = 11632

while True:

RecDraw.clear()

return_value, frame = video_cap.read()

frame_index = frame_index + 1

if frame is None:

break

if frame_index % 2 == 1:

x, y = frame.shape[0:2]

new_image = cv2.resize(frame, (int(y / 2), int(x / 2)))

name += 1

strname = "D:/test/" + str(name) + ".jpg"

cv2.imwrite(strname, new_image)

# transposedImage = cv2.transpose(frame)

# flipedImageX = cv2.flip(transposedImage, 0)

# image_new = Image.fromarray(flipedImageX)

image_new = Image.fromarray(frame)

image, boxes, scores, classes = yolov3.detect_image_mul(image_new)

centers, num = cal_centre(boxes, classes, scores, 0.05)

result = np.asarray(image)

tracker.tracking(centers, num, result, mode)

cv2.namedWindow("result", cv2.WINDOW_NORMAL)

cv2.imshow("result", result)

if isOutput:

out.write(result)

if cv2.waitKey(1) & 0xFF == ord('q'):