No module name 'sklearn.forest.ensemble'

Question

I am using this code to detect face_spoofing

import numpy as np
import cv2
import joblib
from face_detector import get_face_detector, find_faces

def calc_hist(img):
    """
    To calculate histogram of an RGB image

    Parameters
    ----------
    img : Array of uint8
        Image whose histogram is to be calculated

    Returns
    -------
    histogram : np.array
        The required histogram

    """
    histogram = [0] * 3
    for j in range(3):
        histr = cv2.calcHist([img], [j], None, [256], [0, 256])
        histr *= 255.0 / histr.max()
        histogram[j] = histr
    return np.array(histogram)

face_model = get_face_detector()
clf = joblib.load(0)
cap = cv2.VideoCapture("videos/face_spoofing.mp4")

sample_number = 1
count = 0
measures = np.zeros(sample_number, dtype=np.float)

while True:
    ret, img = cap.read()
    faces = find_faces(img, face_model)

    measures[count%sample_number]=0
    height, width = img.shape[:2]
    for x, y, x1, y1 in faces:
        
        roi = img[y:y1, x:x1]
        point = (0,0)
        
        img_ycrcb = cv2.cvtColor(roi, cv2.COLOR_BGR2YCR_CB)
        img_luv = cv2.cvtColor(roi, cv2.COLOR_BGR2LUV)

        ycrcb_hist = calc_hist(img_ycrcb)
        luv_hist = calc_hist(img_luv)

        feature_vector = np.append(ycrcb_hist.ravel(), luv_hist.ravel())
        feature_vector = feature_vector.reshape(1, len(feature_vector))

        prediction = clf.predict_proba(feature_vector)
        prob = prediction[0][1]

        measures[count % sample_number] = prob

        cv2.rectangle(img, (x, y), (x1, y1), (255, 0, 0), 2)

        point = (x, y-5)

        # print (measures, np.mean(measures))
        if 0 not in measures:
            text = "True"
            if np.mean(measures) >= 0.7:
                text = "False"
                font = cv2.FONT_HERSHEY_SIMPLEX
                cv2.putText(img=img, text=text, org=point, fontFace=font, fontScale=0.9, color=(0, 0, 255),
                            thickness=2, lineType=cv2.LINE_AA)
            else:
                font = cv2.FONT_HERSHEY_SIMPLEX
                cv2.putText(img=img, text=text, org=point, fontFace=font, fontScale=0.9,
                            color=(0, 255, 0), thickness=2, lineType=cv2.LINE_AA)
        
    count+=1
    cv2.imshow('img_rgb', img)
    
    if cv2.waitKey(1) & 0xFF == ord('q'):
            break

cap.release()
cv2.destroyAllWindows()

But I get the error

I am using the version 0.24.0 for scikit and am on python 3.8 to use tensorflowTraceback (most recent call last):
  File "C:/Users/heman/PycharmProjects/ProctorAI/face_spoofing.py", line 29, in <module>
    clf = joblib.load('models/face_spoofing.pkl')
  File "C:\Users\heman\PycharmProjects\ProctorAI\venv\lib\site-packages\joblib\numpy_pickle.py", line 585, in load
    obj = _unpickle(fobj, filename, mmap_mode)
  File "C:\Users\heman\PycharmProjects\ProctorAI\venv\lib\site-packages\joblib\numpy_pickle.py", line 504, in _unpickle
    obj = unpickler.load()
  File "C:\Users\heman\AppData\Local\Programs\Python\Python38\lib\pickle.py", line 1212, in load
    dispatch[key[0]](self)
  File "C:\Users\heman\AppData\Local\Programs\Python\Python38\lib\pickle.py", line 1528, in load_global
    klass = self.find_class(module, name)
  File "C:\Users\heman\AppData\Local\Programs\Python\Python38\lib\pickle.py", line 1579, in find_class
    __import__(module, level=0)
ModuleNotFoundError: No module named 'sklearn.ensemble.forest'

Process finished with exit code 1

I think I need to use the previous version of scikit (0.19.1) but I get the error C++ build tools required. I dont know how to install these tools as I am in a virtual environment, they are already installed in my laptop.
Please suggest what I can do

Solution 1

sklearn.ensemble.forest was renamed to sklearn.ensemble._forest in 437ca05 on Oct 16, 2019. You need to install an older sklearn. Try version 0.21.3 released on Jul 30, 2019:

pip install -U scikit-learn==0.21.3

Please be warned that the authors provided wheels up to Python 3.7. For 3.8 or 3.9 you will need to compile from sources.

Solution 2

The answer above is correct, sklearn.ensemble.forest is renamed to sklearn.ensemble._forest

This problem persist with more libraries that depend on sklearn, therefore I want to provide an additional solution that universally for most of these packages.

In your case your library is called face_detector, but you can replace it with any library name when you encounter this issue with versioning of scikit-learn (as well as with other libraries).

1. Locate the directory of the library:

   import face_detector
   print(face_detector.\_\_file__)
   

2. Open the file in any text editor, in your case the name of the library file would be face_detector.py

3. Out-comment the old import and replace with the new import.

   Comment out the import for the old versions of sklearn and add the new import statement

   # from sklearn.ensemble.forest import ForestClassifier, ForestRegressor
   from sklearn.ensemble._forest import ForestClassifier, ForestRegressor
   

4. Safe and enjoy, you just fixed a dependency issue! This solution will work for most libraries and is even less work than installing a different version of sklearn. In case it does not work, you can still install and older version as suggested in the other answer.

Note: This solution can be easily modified to trace and fix dependency issues for other library dependencies than sklearn. As long as the function itself did not change in input and output parameters, fixing renaming issues is an easy way to fix broken dependencies.