opencv/modules/python/test/test_cuda.py
2018-08-07 15:48:12 +09:00

46 lines
1.4 KiB
Python

#!/usr/bin/env python
'''
CUDA-accelerated Computer Vision functions
'''
# Python 2/3 compatibility
from __future__ import print_function
import numpy as np
import cv2 as cv
from tests_common import NewOpenCVTests
class cuda_test(NewOpenCVTests):
def setUp(self):
if not cv.cuda.getCudaEnabledDeviceCount():
self.skipTest("No CUDA-capable device is detected")
def test_cuda_upload_download(self):
npMat = (np.random.random((200, 200, 3)) * 255).astype(np.uint8)
gpuMat = cv.cuda_GpuMat()
gpuMat.upload(npMat)
self.assertTrue(np.allclose(gpuMat.download(), npMat))
def test_cuda_imgproc_cvtColor(self):
npMat = (np.random.random((200, 200, 3)) * 255).astype(np.uint8)
gpuMat = cv.cuda_GpuMat()
gpuMat.upload(npMat)
gpuMat2 = cv.cuda.cvtColor(gpuMat, cv.COLOR_BGR2HSV)
self.assertTrue(np.allclose(gpuMat2.download(), cv.cvtColor(npMat, cv.COLOR_BGR2HSV)))
def test_cuda_filter_laplacian(self):
npMat = (np.random.random((200, 200)) * 255).astype(np.uint16)
gpuMat = cv.cuda_GpuMat()
gpuMat.upload(npMat)
gpuMat = cv.cuda.createLaplacianFilter(cv.CV_16UC1, -1, ksize=3).apply(gpuMat)
self.assertTrue(np.allclose(gpuMat.download(), cv.Laplacian(npMat, cv.CV_16UC1, ksize=3)))
if __name__ == '__main__':
NewOpenCVTests.bootstrap()