5 月 7 日广东韶关一辆特斯拉 Model 3 追尾一辆货车，目前情况如何？事故原因可能是怎样的？ 第1页

       def video_to_pic(input_file,output_dir):     cap = cv2.VideoCapture(input_file)     cnt = 0     if(not os.path.exists(output_dir)):         os.makedirs(output_dir)          ret, frame = cap.read()     while ret:         cv2.imwrite(output_dir+"/{0:04d}.jpg".format(cnt), frame)         ret, frame = cap.read()         cnt += 1     

       def pic_normalize(pic_file,target_file,out_file):     pic1 = cv2.imread(target_file)     pic2 = cv2.imread(pic_file)          sift=cv2.xfeatures2d.SIFT_create()     k1, d1 = sift.detectAndCompute(pic1, None)     k2, d2 = sift.detectAndCompute(pic2, None)     matcher = cv2.BFMatcher()     mat = matcher.knnMatch(d1, d2, k=2)     fmat = []     for m1, m2 in mat:         if m1.distance < 0.8 * m2.distance:             fmat.append((m1.trainIdx, m1.queryIdx))     km1 = np.float32([k1[i].pt for (t, i) in fmat])     km2 = np.float32([k2[i].pt for (i, t) in fmat])     H, t = cv2.findHomography(km2, km1, cv2.RANSAC,5.0)          rst = cv2.warpPerspective(pic2, H, (pic2.shape[1], pic2.shape[0]))     cv2.imwrite(out_file, rst)  def normalize(begin_num,end_num,input_dir,output_dir):     cnt = 0     if(not os.path.exists(output_dir)):         os.makedirs(output_dir)     for i in range(begin_num,end_num+1):         pic_normalize(             input_dir+"/{0:04d}.jpg".format(i),             input_dir+"/{0:04d}.jpg".format(end_num),             output_dir+"/{0:04d}.jpg".format(cnt)         )         cnt += 1     

       def labelize(begin_num,end_num,input_dir,output_dir):     global cnt     if(not os.path.exists(output_dir)):         os.makedirs(output_dir)     rst=[]     cnt=0     pic=cv2.imread(input_dir+"/{0:04d}.jpg".format(cnt))     def mouse_down(event, x, y, flags, param):         global cnt         if event == cv2.EVENT_LBUTTONDOWN:             rst.append((x,y))                         pic=cv2.imread(input_dir+"/{0:04d}.jpg".format(cnt))             cv2.circle(pic, rst[-1], 3, (0, 0, 255), thickness = -1)             cv2.imwrite(output_dir+"/{0:04d}.jpg".format(cnt), pic)             cnt+=1             if(cnt==end_num+1):                 cv2.destroyAllWindows()                 return             pic=cv2.imread(input_dir+"/{0:04d}.jpg".format(cnt))             cv2.imshow("image", pic)          cv2.namedWindow("image")     cv2.setMouseCallback("image", mouse_down)     cv2.imshow("image", pic)     cv2.waitKey(0)              return rst     

       def calcAndPlot(rst,ratio):     xrst=[i[0] for i in rst]     yrst=[i[1] for i in rst]          xfrst=[]     yfrst=[]          for i in range(len(rst)-4):         xfrst.append(sum(xrst[i:i+5])/5)         yfrst.append(sum(yrst[i:i+5])/5)          v=[]     for i in range(len(xfrst)-1):         v.append(((xfrst[i]-xfrst[i+1])**2+(yfrst[i]-yfrst[i+1])**2)**0.5)          v=[i*ratio for i in v]              x=[i/24 for i in range(len(v))]     plt.plot(x,v)     plt.xlabel("time(second)")     plt.ylabel("speed(kmph)")     plt.ylim((0,200))     plt.show()     return v