System Information

I'm using UMat type frames to make use of my Intel UHD Graphics 620 and I have discovered a bug in WarpAffine function returning unexpected results.

I have opened a blog post on forum for more information.

OpenCV python version: 4.10.0
Operating System / Platform: Windows 10
Python version: 3.12.2

Detailed description

As you can see in my following code I'm using same UMat for input and output.
I have tried using new UMatfor output and seems it works.

Note: I have discovered the bug using EmguCv because I'm working on C#. But to be sure I have tested the issue in Python to reproduce it directly on OpenCv.

Steps to reproduce

Here is an example Python code to test.

import os
os.add_dll_directory("C:\\gstreamer\\1.0\\msvc_x86_64\\bin")
 
import cv2 as cv
 
print(cv.getBuildInformation())
 
img_srcFile = "C:\\Users\\usr\\Downloads\\demo.jpg"
 
srcMat = cv.imread(img_srcFile, cv.IMREAD_UNCHANGED)
srcUMat = cv.UMat(cv.imread(img_srcFile, cv.IMREAD_UNCHANGED))
 
cv.imshow("SrcMat", srcMat)
cv.imshow("SrcUMat", srcUMat)
 
height, width = srcMat.shape[:2]
 
point = ((width - 1 )/ 2, (height - 1 )/ 2)
matrixTransfrom = cv.getRotationMatrix2D(point, 0, 1.3)
 
cv.warpAffine(srcMat, matrixTransfrom, (width, height), srcMat)
cv.imshow("SrcMat -> Zoom", srcMat)
 
cv.warpAffine(srcUMat, matrixTransfrom, (width, height), srcUMat)
cv.imshow("SrcUMat -> Zoom", srcUMat)
 
cv.waitKey(0)

I have fixed temporally using different UMat for output.

dstImg = cv.warpAffine(srcUMat, matrixTransfrom, (width, height))
cv.imshow("SrcUMat -> Zoom", dstImg)

Issue submission checklist

• I report the issue, it's not a question
• I checked the problem with documentation, FAQ, open issues, forum.opencv.org, Stack Overflow, etc and have not found any solution
• I updated to the latest OpenCV version and the issue is still there
• There is reproducer code and related data files (videos, images, onnx, etc)