如何将 <class 'streamlit.uploaded_file_manager.UploadedFile'> 类型的文件转换为 Ptr<cv::UMat>？

Posted 2023-03-29

【中文标题】如何将 <class \'streamlit.uploaded_file_manager.UploadedFile\'> 类型的文件转换为 Ptr<cv::UMat>？

【英文标题】：How to convert a file of type <class 'streamlit.uploaded_file_manager.UploadedFile'> to Ptr<cv::UMat>?如何将 <class 'streamlit.uploaded_file_manager.UploadedFile'> 类型的文件转换为 Ptr<cv::UMat>？

【发布时间】：2021-08-07 10:15:33

【问题描述】：

我想要做的是通过 Streamlit file_uploader 上传一个文件，然后将该 png 传递给一个函数进行预测。这是我的代码的样子：

if choice == "Upload Image":
    image_file = st.sidebar.file_uploader('Upload an image', type = 'png')
    if image_file and st.sidebar.button('Load'):
        image = get_opened_image(image_file)
        with st.beta_expander('Selected Image', expanded=True):
            st.image(image, use_column_width=True)
            prediction = image_pred(image_file)   # line 52
            st.subheader("Prediction")
            st.markdown(f'The predicted label is: **prediction**')

这是调用的预测函数：

def image_pred(filename):
    emo = []
    model = model_from_json(open("fer.json", "r").read())
    model.load_weights('fer.h5')
    emotion_dict = 0: "Not Engaged", 1: "Not Engaged", 2: "Nominally Engaged", 3: "Very Engaged", 4: "Not Engaged",
                    5: "Very Engaged", 6: "Nominally Engaged"
    facecasc = cv2.CascadeClassifier(cv2.data.haarcascades + 'haarcascade_frontalface_default.xml')
    gray = cv2.cvtColor(filename, cv2.COLOR_BGR2GRAY) # line 29 error here
    faces = facecasc.detectMultiScale(gray, scaleFactor=1.3, minNeighbors=5)


 

这是我得到的错误：

TypeError: Expected Ptr<cv::UMat> for argument 'src'

line 52, in <module>
    prediction = image_pred(image_file)

line 29, in image_pred
    gray = cv2.cvtColor(filename, cv2.COLOR_BGR2GRAY)

如何将上传的文件转换为可以作为 src 传递给 cv2.cvtColor() 的合适格式？提前致谢

【参考方案1】：

cv2.cvtColor() 需要图像数据作为第一个参数，而不是文件名。你可以这样得到：

image = cv2.imread(filename)
# now convert the color space
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)