使用OpenCV编写视频时出现错误的视频分类

我正在构建一个与此视频中相同的视频分类模型:https://www.youtube.com/watch?v=l8XQsxlGxiY&list=PLxefhmF0pcPl_v-lLsqF3drP6NOoSYJR0&index=9

该模型预测视频中展示的是哪项运动:网球、游泳或拳击。

我已经构建了我的模型,准确率达到了96%。现在,我有一个样本视频来测试我的模型。

from keras.models import load_modelfrom collections import dequeimport numpy as npimport pickleimport cv2
model = load_model(r"C:\Users\yudishteer.c\Desktop\VideoClassification\video_classification_model\videoclassificationModel")lb = pickle.loads(open(r"C:\Users\yudishteer.c\Desktop\VideoClassification\video_classification_model\videoclassificationBinarizer.pickle", "rb").read())outputvideo = r"C:\Users\yudishteer.c\Desktop\VideoClassification\video_classification_model\outputvideo\demo_output.avi"mean = np.array([123.68, 116.779, 103.939], dtype = "float32")Queue = deque(maxlen=128)
capture_video=cv2.VideoCapture(r"C:\Users\yudishteer.c\Desktop\VideoClassification\video_classification_model\demo_video.mp4")writer = None(Width, Height) = (None, None)while True:    (taken, frame) = capture_video.read()    if not taken:        break    if Width is None or Height is None:        (Width, Height) = frame.shape[:2]            output = frame.copy()    frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)    frame = cv2.resize(frame, (224,224)).astype("float32")    frame -= mean    preds = model.predict(np.expand_dims(frame, axis=0))[0]    Queue.append(preds)    results = np.array(Queue).mean(axis = 0)    i = np.argmax(results)    label = lb.classes_[i]    text = "The game is {}".format(label)    cv2.putText(output, text, (45,60), cv2.FONT_HERSHEY_SIMPLEX, 1.25, (255,0,0),5)        if writer is None:        fourcc = cv2.VideoWriter_fourcc(*'MJPG')        writer = cv2.VideoWriter('demo_output.mp4', fourcc, 10, (Width, Height), True)    writer.write(output)    cv2.imshow("In progress", output)    key = cv2.waitKey(1) & 0xFF        if key == ord("q"):        break        print("Finalizing.......")writer.release()capture_video.release()# Closes all the framescv2.destroyAllWindows()

当我运行上述代码时,我可以看到它被正确分类。也就是说,演示视频(demo_video.mp4)被打开,我可以看到根据展示的运动,视频顶部显示的是游泳、网球或拳击。

然而,由于我有以下代码:writer = cv2.VideoWriter('demo_output.mp4', fourcc, 10, (Width, Height), True)

当我打开demo_output.mp4(甚至是avi)时,我得到了以下错误:

error

有人能帮我找出问题所在吗?谢谢!


回答:

你混淆了WidthHeight

不是(Width, Height) = frame.shape[:2],应该是:

(Height, Width) = frame.shape[:2]

在我的系统中,我收到了以下警告:

OpenCV: FFMPEG: tag 0x47504a4d/’MJPG’ is not supported with codec id 7 and format ‘mp4 / MP4 (MPEG-4 Part 14)’OpenCV: FFMPEG: fallback to use tag 0x7634706d/’mp4v’

结果是一个黑屏视频。

以下编码器/容器组合之一可以工作:

  • 'MJPG' FOURCC 和 AVI 容器:

     fourcc = cv2.VideoWriter_fourcc(*'MJPG') writer = cv2.VideoWriter('demo_output.avi', fourcc, 10, (Width, Height), True)
  • 'mp4v' FOURCC 和 MP4 容器:

     fourcc = cv2.VideoWriter_fourcc(*'mp4v') writer = cv2.VideoWriter('demo_output.mp4', fourcc, 10, (Width, Height), True)

Related Posts

使用LSTM在Python中预测未来值

这段代码可以预测指定股票的当前日期之前的值,但不能预测…

如何在gensim的word2vec模型中查找双词组的相似性

我有一个word2vec模型,假设我使用的是googl…

dask_xgboost.predict 可以工作但无法显示 – 数据必须是一维的

我试图使用 XGBoost 创建模型。 看起来我成功地…

ML Tuning – Cross Validation in Spark

我在https://spark.apache.org/…

如何在React JS中使用fetch从REST API获取预测

我正在开发一个应用程序,其中Flask REST AP…

如何分析ML.NET中多类分类预测得分数组?

我在ML.NET中创建了一个多类分类项目。该项目可以对…

发表回复

您的邮箱地址不会被公开。 必填项已用 * 标注