安装Imageai2.1.5后无法从"BatchNormalization“导入名称"keras.layers.normalization”

Question

我试过安装火把，但失败了。我还下载了运行它所需的所有库(opencv-python、numpy、scipy等)，但没有.我怎样才能解决这个问题？

错误：

File "C:\Users\Home\PycharmProjects\XG -I programming(trial part 2)\Mask_detection(trial 6).py", line 1, in <module>
    from imageai.Detection import VideoObjectDetection
  File "C:\Users\Home\PycharmProjects\XG -I programming(trial part 2)\venv\lib\site-packages\imageai\Detection\__init__.py", line 17, in <module>
    from imageai.Detection.YOLOv3.models import yolo_main, tiny_yolo_main
  File "C:\Users\Home\PycharmProjects\XG -I programming(trial part 2)\venv\lib\site-packages\imageai\Detection\YOLOv3\models.py", line 8, in <module>
    from keras.layers.normalization import BatchNormalization 
ImportError: cannot import name 'BatchNormalization' from 'keras.layers.normalization' (C:\Users\Home\PycharmProjects\XG -I programming(trial part 2)\venv\lib\site-packages\keras\layers\normalization\__init__.py)

代码：

from imageai.Detection import VideoObjectDetection
import cv2

camera = cv2.VideoCapture(0)
detector = VideoObjectDetection()

model_path = "./Models/yolo.h5"

detector.setModelTypeAsYOLOv3()
detector.setModelPath(model_path)
detector.loadModel()

while True:
ret, frame = camera.read()

rgb_frame = frame[:, :, ::-1]

video_detection = detector.detectObjectsFromVideo(camera_input=camera, output_file_path ="./camera_detected_video", frames_per_second=28, log_progress=True, minimum_percentage_probability=40)
cv2.imshow('frame', frame)

print(video_detection)

if cv2.waitKey(1) & 0xff == ord('q'):
    break


camera.release()
cv2.destroyAllWindows()

Answer 1

问题在于keras版本，请尝试安装旧的keras版本。

安装pip3 keras==2.4.3