Untitled

int main() {
    // Load an RGB image using OpenCV
    cv::Mat inputImage = cv::imread("/home/hesam/Desktop/playground/cuda/DSC_0325.JPG", cv::IMREAD_COLOR);
    cv::resize(inputImage, inputImage, cv::Size(600, 400));
    if (inputImage.empty()) {
        std::cerr << "Error: Could not load the image." << std::endl;
        return -1;
    }

    int width = inputImage.cols;
    int height = inputImage.rows;
    size_t imageSize = width * height * sizeof(uchar3);

    // Allocate Unified Memory for image data
    uchar3* unifiedInputImage;
    uchar3* unifiedOutputImage;
    cv::Mat* unifiedResultImage; // Pointer to the final result in Unified Memory

    cudaMallocManaged(&unifiedInputImage, imageSize);
    cudaMallocManaged(&unifiedOutputImage, imageSize);
    cudaMallocManaged(&unifiedResultImage, sizeof(cv::Mat));

    // Load the input image directly into Unified Memory
    memcpy(unifiedInputImage, inputImage.ptr<uchar3>(), imageSize);

    // Define block and grid dimensions
    dim3 blockSize(16, 16);
    dim3 gridSize((width + blockSize.x - 1) / blockSize.x, (height + blockSize.y - 1) / blockSize.y);

    // Launch the convolution kernel on the GPU
    convolutionKernelGPU<<<gridSize, blockSize>>>(unifiedInputImage, unifiedOutputImage, width, height);
    cudaDeviceSynchronize();

    *unifiedResultImage = cv::Mat(height, width, CV_8UC3, unifiedOutputImage);

    // Save the resulting image
    cv::imwrite("blurred.jpeg", *unifiedResultImage);

    // Free Unified Memory
    cudaFree(unifiedInputImage);
    cudaFree(unifiedOutputImage);
    cudaFree(unifiedResultImage);

    return 0;
}