Question

我是OpenCV python以及StackOverflow的初学者，所以请原谅我任何错误。我正在使用OpenCV 3.2和Python 3.6进行图像分类任务。我的任务是对帧中的对象进行分类，并在该帧上给出其类标签。当写回最终输出视频时，放在视频上的文本会重叠。下面是我的代码。

cap = cv2.VideoCapture(r"D:\python\tank_video\All_vehicle.mp4")
length = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
currentFrame = 0

# videowriter object
codc = int(cap.get(cv2.CAP_PROP_FOURCC))
fps = int(cap.get(cv2.CAP_PROP_FPS))
w = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
h = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))

fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter('classify_300_200_3.avi', fourcc, 15, (300,200))

                            "CLASSIFICATION CODE"

predictions =  [classes_names[i] for i in clf.predict(test_features)]
arr = np.array([])

j = 0
for prediction in predictions:
    pt = (0, 3 * image.shape[0] // 4)
    cv2.putText(image, prediction, pt ,cv2.FONT_HERSHEY_SIMPLEX, 2, [0, 255, 
    0], 2)
    np.append(arr,cv2.imwrite(str(j)+'.jpg',image))

out.write(image)
currentFrame += 1
cap.release()        
out.release()
cv2.destroyAllWindows()

输出图像已附在此处。this is the screenshot of video output.我想要显示一个类标签。不应显示前一帧的标签，并且不应在前一帧的标签上覆盖当前帧的标签。任何人都可以帮我解决吗？

Answer 1

如果您只想显示一个类标签，则应更改以下内容：

for prediction in predictions:
    pt = (0, 3 * image.shape[0] // 4)
    cv2.putText(image, prediction, pt ,cv2.FONT_HERSHEY_SIMPLEX, 2, [0, 255, 
    0], 2)
    np.append(arr,cv2.imwrite(str(j)+'.jpg',image))

因为那段代码会将检测到的所有类写入图像。例如，你可以这样做：

# check if at least one class detected
if len(predictions)>0:
    # choose to display only the first class
    prediction = predictions[0]

    pt = (0, 3 * image.shape[0] // 4)
    cv2.putText(image, prediction, pt ,cv2.FONT_HERSHEY_SIMPLEX, 2, [0, 255, 
    0], 2)
    np.append(arr,cv2.imwrite(str(j)+'.jpg',image))

Opencv 3.2 Python视频写作问题：帧上的文本重叠

1 个答案: