def __init__(self):
# Init pygame
pygame.init()
# Creat pygame window
pygame.display.set_caption("Tello video stream")
self.screen = pygame.display.set_mode([960, 720])
# Init Tello object that interacts with the Tello drone
self.tello = Tello()
# Drone velocities between -100~100
self.for_back_velocity = 0
self.left_right_velocity = 0
self.up_down_velocity = 0
self.yaw_velocity = 0
self.speed = 10
self.mode = None
self.send_rc_control = False
self.yolo = Yolo()
self.yolo.initializeModel()
self.tracker = tracker = cv2.TrackerCSRT().create()
self.locked = False
self.locked_frame = None
# create update timer
pygame.time.set_timer(USEREVENT + 1, 50)
logger.info("Game Initialized")
def playFirstCamera(self,detection_graph,sess,img):
# 显示轮询法参数
cv2.putText(img, str(self.firstTimes), (0, 20), cv2.FONT_HERSHEY_SIMPLEX, 0.75,
(0, 0, 255), 2)
# cv2.rectangle(img, (340, 450), (800, 700), (205, 240, 0), 2)
if self.firstTimes!=0:
self.firstTimes-=1
else:
image_np_expanded = np.expand_dims(img, axis=0)
height,width, a = img.shape
image_tensor = detection_graph.get_tensor_by_name('image_tensor:0')
# Each box represents a part of the image where a particular object was detected.
boxes = detection_graph.get_tensor_by_name('detection_boxes:0')
# Each score represent how level of confidence for each of the objects.
# Score is shown on the result image, together with the class label.
scores = detection_graph.get_tensor_by_name('detection_scores:0')
classes = detection_graph.get_tensor_by_name('detection_classes:0')
num_detections = detection_graph.get_tensor_by_name('num_detections:0')
# Actual detection.
(boxes, scores, classes, num_detections) = sess.run(
[boxes, scores, classes, num_detections],
feed_dict={image_tensor: image_np_expanded})
# vis_util.visualize_boxes_and_labels_on_image_array(
# img,
# np.squeeze(boxes),
# np.squeeze(classes).astype(np.int32),
# np.squeeze(scores),
# category_index,
# use_normalized_coordinates=True,
# line_thickness=8,
# min_score_thresh=0.9)
#过滤置信度低于80%的目标
s_boxes = boxes[scores > 0.8]
for i in range(0, s_boxes.shape[0]):
ymin = int(s_boxes[i][0] * height) # ymin
xmin = int(s_boxes[i][1] * width) # xmin
ymax = int(s_boxes[i][2] * height) # ymax
xmax = int(s_boxes[i][3] * width) # xmax
# print(str(xmin)+" "+str(xmax)+" "+str(ymin)+" "+str(ymax))
# if xmax>xmin+1 and ymax>ymin+1:
#设置监控范围在【450，700】的竖直方向
if ymin >= 450 and ymax <= 700:
flag = True
midx = int(xmin + (xmax - xmin) / 2)
midy = int(ymin + (ymax - ymin) / 2)
for i in range(len(self.trackersPositionSecond)):
if self.trackersPositionSecond[i][0] < midx < self.trackersPositionSecond[i][2] and self.trackersPositionSecond[i][1] < midy < \
self.trackersPositionSecond[i][3]:
flag = False
break
if flag:
tracker = cv2.TrackerCSRT().create()
# tracker = cv2.TrackerKCF().create()
# tracker = cv2.TrackerMOSSE().create()
tracker.init(img, (xmin, ymin, xmax - xmin, ymax - ymin))
self.trackersSecond.append(tracker)
self.trackersPositionSecond.append((xmin, ymin, xmax, ymax))
self.firstTimes = 15
out = queue.Queue()
# 更新跟踪器
for i in range(len(self.trackersSecond)):
tracker = self.trackersSecond[i]
ok, bboxes = tracker.update(img)
if ok:
p1 = (int(bboxes[0]), int(bboxes[1]))
p2 = (int(bboxes[0] + bboxes[2]), int(bboxes[1] + bboxes[3]))
self.trackersPositionSecond[i] = (
int(bboxes[0]), int(bboxes[1]), int(bboxes[0] + bboxes[2]), int(bboxes[1] + bboxes[3]))
# cv2.rectangle(img, p1, p2, (0, 0, 255), 2, 10)
#判断是否到达结束区域
if int(bboxes[1] + bboxes[3]) < 550:
ymin = int(bboxes[1]) - 15
xmin = int(bboxes[0]) - 15
ymax = int(bboxes[1] + bboxes[3]) + 15
xmax = int(bboxes[0] + bboxes[2]) + 15
part = img[ymin:ymax, xmin:xmax]
self.listWaitToDetecte.put((part, 1))
# result = self.detecte2(detection_graph,sess,part)
# if result is None:
# print("有行李未被检测出")
# else:
# cv2.imwrite(r"./img/second/" + str(time.time()) + ".jpg", result)
out.put(i)
if int(bboxes[2]) * int(bboxes[3]) >= 150000:
print("有行李出现错误，失效")
out.put(i)
while not out.empty():
outIndex = out.get()
self.trackersSecond.pop(outIndex)
self.trackersPositionSecond.pop(outIndex)
return img
def playThirdCamera(self,detection_graph,sess,img):
# 显示轮询法参数
cv2.putText(img, str(self.thirdTimes), (0, 20), cv2.FONT_HERSHEY_SIMPLEX, 0.75,
(0, 0, 255), 2)
# cv2.rectangle(img, (580, 370), (1050, 680), (205, 240, 0), 2)
if self.thirdTimes!=0:
self.thirdTimes-=1
else:
image_np_expanded = np.expand_dims(img, axis=0)
height, width, a = img.shape
image_tensor = detection_graph.get_tensor_by_name('image_tensor:0')
# Each box represents a part of the image where a particular object was detected.
boxes = detection_graph.get_tensor_by_name('detection_boxes:0')
# Each score represent how level of confidence for each of the objects.
# Score is shown on the result image, together with the class label.
scores = detection_graph.get_tensor_by_name('detection_scores:0')
classes = detection_graph.get_tensor_by_name('detection_classes:0')
num_detections = detection_graph.get_tensor_by_name('num_detections:0')
# Actual detection.
(boxes, scores, classes, num_detections) = sess.run(
[boxes, scores, classes, num_detections],
feed_dict={image_tensor: image_np_expanded})
# 可视化工具
# vis_util.visualize_boxes_and_labels_on_image_array(
# img,
# np.squeeze(boxes),
# np.squeeze(classes).astype(np.int32),
# np.squeeze(scores),
# category_index,
# use_normalized_coordinates=True,
# line_thickness=8,
# min_score_thresh=0.9)
#过滤置信度低于80%的行李
s_boxes = boxes[scores > 0.8]
for i in range(0, s_boxes.shape[0]):
ymin = int(s_boxes[i][0] * height) # ymin
xmin = int(s_boxes[i][1] * width) # xmin
ymax = int(s_boxes[i][2] * height) # ymax
xmax = int(s_boxes[i][3] * width) # xmax
# print(str(xmin)+" "+str(xmax)+" "+str(ymin)+" "+str(ymax))
# if xmax>xmin+1 and ymax>ymin+1:
# 设置监控区域范围 ，在[370,680]竖直区间内
if ymin >= 370 and ymax <= 680:
flag = True
midx = int(xmin + (xmax - xmin) / 2)
midy = int(ymin + (ymax - ymin) / 2)
for i in range(len(self.trackersPositionThird)):
if self.trackersPositionThird[i][0] < midx < self.trackersPositionThird[i][2] and \
self.trackersPositionThird[i][1] < midy < \
self.trackersPositionThird[i][3]:
flag = False
break
if flag:
# 初始化追踪器
tracker = cv2.TrackerCSRT().create()
# tracker = cv2.TrackerKCF().create()
# tracker = cv2.TrackerMOSSE().create()
tracker.init(img, (xmin, ymin, xmax - xmin, ymax - ymin))
self.trackersThird.append(tracker)
self.trackersPositionThird.append((xmin, ymin, xmax, ymax))
self.thirdTimes = 20
out = queue.Queue()
# 跟新追踪器
for i in range(len(self.trackersThird)):
tracker = self.trackersThird[i]
ok, bboxes = tracker.update(img)
if ok:
p1 = (int(bboxes[0]), int(bboxes[1]))
p2 = (int(bboxes[0] + bboxes[2]), int(bboxes[1] + bboxes[3]))
self.trackersPositionThird[i] = (
int(bboxes[0]), int(bboxes[1]), int(bboxes[0] + bboxes[2]), int(bboxes[1] + bboxes[3]))
# cv2.rectangle(img, p1, p2, (0, 0, 255), 2, 10)
# 判断是否到达结束区域
if int(bboxes[1] + bboxes[3]) < 470:
ymin = int(bboxes[1]) - 20
xmin = int(bboxes[0]) - 20
ymax = int(bboxes[1] + bboxes[3]) + 20
xmax = int(bboxes[0] + bboxes[2]) + 20
part = img[ymin:ymax, xmin:xmax]
self.listWaitToDetecte.put((part, 3))
out.put(i)
# 判断面积，当面积太大时，说明追踪算法已经失效，应该舍弃
if int(bboxes[2])*int(bboxes[3])>=150000:
print("有行李出现错误，失效")
out.put(i)
while not out.empty():
outIndex = out.get()
self.trackersThird.pop(outIndex)
self.trackersPositionThird.pop(outIndex)
return img
# image Processing
import numpy as np
import cv2
import numpy as np
import cv2 as cv
from ImageProcessing.yolov3 import Yolo
face_cascade = cv.CascadeClassifier("haarcascade_frontalface_default.xml")
eye_cascade = cv.CascadeClassifier("haarcascade_eye.xml")
cap = cv2.VideoCapture(0)
yolo = Yolo()
yolo.initializeModel()
tracker = cv2.TrackerCSRT().create()
tracker_start = False
first_frame = True
# size=
out = cv2.VideoWriter("project.avi", cv2.VideoWriter_fourcc(*"DIVX"), 15,
(1280, 720))
while True:
# Capture frame-by-frame
ret, frame = cap.read()
shape = frame.shape
print(shape)
fx = 0.5
fy = 0.5
# Our operations on the frame come here
img = cv2.resize(frame, None, fx=0.5, fy=0.5)