快捷導(dǎo)航

對(duì)YOLOv3模型調(diào)用時(shí)候的python接口詳解

更新時(shí)間：2019年08月26日 11:26:58 作者：NcepuKZH

今天小編就為大家分享一篇對(duì)YOLOv3模型調(diào)用時(shí)候的python接口詳解，具有很好的參考價(jià)值，希望對(duì)大家有所幫助。一起跟隨小編過(guò)來(lái)看看吧

需要注意的是：更改完源程序.c文件，需要對(duì)整個(gè)項(xiàng)目重新編譯、make install，對(duì)已經(jīng)生成的文件進(jìn)行更新，類似于之前VS中在一個(gè)類中增加新函數(shù)重新編譯封裝dll，而python接口的調(diào)用主要使用的是libdarknet.so文件，其余在配置文件中的修改不必重新進(jìn)行編譯安裝。

之前訓(xùn)練好的模型，在模型調(diào)用的時(shí)候，總是在

lib = CDLL("/home/*****/*******/darknet/libdarknet.so", RTLD_GLOBAL)這里讀不到darknet編譯生成的.so文件，導(dǎo)致直接的報(bào)錯(cuò)；之前以為是文件路徑的問(wèn)題，稀里糊涂的；由于很久不寫c文件了，所以最后直接在python接口后在py文件中修改的畫框、標(biāo)置信度等操作，一次次的嘗試后終于成功

（1）將項(xiàng)目中python文件下的darknet.py文件拷貝到根目錄，和/libdarknet.so在同一個(gè)目錄下

（2）整個(gè)demo程序都是用絕對(duì)路徑；

實(shí)現(xiàn)yolov3模型加載，批量讀取文件夾下的照片到庫(kù)函數(shù)變量，最終處理結(jié)果存入在另外新建文件夾

###2019.04.03 by ylxb
def showPicResult(image,peoplecar，outimage):
  img = cv2.imread(image)
  out_img =outimage
  cv2.imwrite(out_img, img)
  for i in range(len(peoplecar)):
    x1=peoplecar[i][2][0]-peoplecar[i][2][2]/2
    y1=peoplecar[i][2][1]-peoplecar[i][2][3]/2
    x2=peoplecar[i][2][0]+peoplecar[i][2][2]/2
    y2=peoplecar[i][2][1]+peoplecar[i][2][3]/2
    im = cv2.imread(out_img)
    cv2.rectangle(im,(int(x1),int(y1)),(int(x2),int(y2)),(255,255,0),3)
    text = listpeoplecar[i][0]
    # 在圖片上添加文字信息
    if(text=="people"):
      carcol=(55, 55, 255)#顏色顯示
    else:
      carcol = (255, 55, 55)
    cv2.putText(im, text, (int(x1), int(y1)), cv2.FONT_HERSHEY_SIMPLEX,
          0.8, carcol, 1, cv2.LINE_AA)
    #This is a method that works well.
    cv2.imwrite(out_img, im)

###2019.04.03 by ylxb
  filenames = os.listdir(picDir)
  i = 0
  num = 0#目標(biāo)個(gè)數(shù)
  car_num = 0#car個(gè)數(shù)
  people_num = 0#people個(gè)數(shù)

  car = "car" # car元素
  people = "people" # people元素

  for name in filenames:
    filename=os.path.join(picDir,name)
    #print(filename)
    listpeoplecar = detect(net, meta, filename)
    print(listpeoplecar)
    i = i + 1
    #save_picpath = out_img+str(filename).split("/")[-1].split(".")[0] + ".png"
    out_img=out_img1+str(i)+'.png'
    showPicResult(filename,listpeoplecar，out_img)

    for item in listpeoplecar:
      #print(item)
      car_num = car_num + item[0].count(car)#car個(gè)數(shù)
      people_num = people_num + item[0].count(people)#people個(gè)數(shù)
      num = num + 1#目標(biāo)個(gè)數(shù)

  print('car個(gè)數(shù): ' + str(car_num))
  print('people個(gè)數(shù): ' + str(people_num))
  print('共檢測(cè)出目標(biāo)個(gè)數(shù): ' + str(num))
  print('共檢測(cè)照片個(gè)數(shù):'+ str(i))

放其中一個(gè)照片測(cè)試照片：