Python创建SIFT遇AttributeError：模块无SIFT_create属性求助

Hey there! Let's break down why you're getting that AttributeError: module 'cv2.cv2' has no attribute 'SIFT_create' error when running your feature matching code.

The Root Cause: OpenCV Version & SIFT's Module Placement #

SIFT (Scale-Invariant Feature Transform) had patent restrictions until 2020, which changed where it's located in OpenCV across versions:

• OpenCV 2.x: You’d use cv2.SIFT() directly (this is what your original code tried, but this syntax is long deprecated).
• OpenCV 3.x to 4.3: SIFT was moved to the opencv-contrib-python package (extra contributed modules), so you had to call it via cv2.xfeatures2d.SIFT_create().
• OpenCV 4.4+ (post-patent expiration): SIFT was moved back to the main OpenCV module, so cv2.SIFT_create() works—but only if you have the full package installed.

Step-by-Step Fixes #

1. Install/Update the Right OpenCV Packages
   If you only installed opencv-python, you’re missing the contrib modules (required for SIFT in older versions, and useful for other features). Run this in your terminal to get both core and contrib packages:

   pip install --upgrade opencv-python opencv-contrib-python
   

2. Check Your OpenCV Version
   First, confirm which version you’re using to pick the correct SIFT initialization:

   import cv2
   print(cv2.__version__)
   

   • For 4.4 or newer: Use cv2.SIFT_create()
   • For 3.x to 4.3: Use cv2.xfeatures2d.SIFT_create()

3. Full Corrected Code Example
   Here’s a version that works across most modern OpenCV releases, with a fallback for older versions:

   import numpy as np
   import cv2
   from matplotlib import pyplot as plt
   
   img1 = cv2.imread('or1.jpg', 0)
   img2 = cv2.imread('dm1.jpg', 0)
   
   # Initialize SIFT with version fallback
   try:
       sift = cv2.SIFT_create()
   except AttributeError:
       sift = cv2.xfeatures2d.SIFT_create()
   
   kp1, des1 = sift.detectAndCompute(img1, None)
   kp2, des2 = sift.detectAndCompute(img2, None)
   
   bf = cv2.BFMatcher()
   matches = bf.knnMatch(des1, des2, k=2)
   
   good = []
   for m, n in matches:
       if m.distance < 0.75 * n.distance:
           good.append([m])
   
   # Added None as the output image parameter to match modern syntax
   img3 = cv2.drawMatchesKnn(img1, kp1, img2, kp2, good, None, flags=2)
   plt.imshow(img3), plt.show()
   

Why Your Original Code Failed #

• Your first attempt used sift = cv2.SIFT()—this is outdated syntax that only works in very old OpenCV 2.x versions.
• When you switched to cv2.SIFT_create(), if you didn’t have the correct OpenCV package (either missing contrib for older versions, or an outdated main package), Python couldn’t locate the function, hence the AttributeError.

内容的提问来源于stack exchange，提问作者Humair Mannan