nvjpeg-python icon indicating copy to clipboard operation
nvjpeg-python copied to clipboard

nvjpeg for python

Results 18 nvjpeg-python issues
Sort by recently updated
recently updated
newest added
trafficstars

jetson@jetson-orin-nx:~/nvjpeg/nvjpeg-python$ pip install pynvjpeg Defaulting to user installation because normal site-packages is not writeable Collecting pynvjpeg Using cached pynvjpeg-0.0.13.tar.gz (11 kB) Preparing metadata (setup.py) ... done Requirement already satisfied: numpy>=1.17...

我分别在RTX 4090和Jetson TX2 NX的虚拟环境中使用pip install pynvjpeg安装了pynvjpeg。 在RTX 4090和Jetson TX2 NX中,我分别使用OpenCV中的imencode和pynvjpeg中的encode方法对分辨率4K的图片进行编码时间测试。 在RTX 4090中,使用OpenCV的编码方式,编码时间70ms;使用pynvjpeg的编码方式,编码时间10ms 在Jetson中,使用OpenCV的编码方式,编码时间110ms;使用pynvjpeg的编码方式,编码时间150ms 为什么在Jetson中,使用pynvjpeg的编码方式,编码时间会更长?

不知道为什么,把下面函数返回的 unsigned char* 值给cv::Mat之后会导致内存泄漏 ``` unsigned char* JpegCoderImage::buffer() { nvjpegImage_t* img = ((nvjpegImage_t*)(this->img)); size_t size = height * width * 3; unsigned char* buffer = (unsigned char*)malloc(size); cudaMemcpy(buffer, img->channel[0], size,...

总体来说谢谢您开源这么好的库,非常棒的工作-------爱来自瓷器 祝您长生不老,永远不死,财源广进,事事顺心。 您就是活菩萨。

Hello there! I'm trying to do some speed tests writting encoded images to disk playing with jpeg quality factor. I'm using the python wrapper for NvJpeg_write as follows: ``` foo...

现在得把cuda和NvJpeg都安装到anaconda里方便复制环境。 cuda可以通过conda安装,直接都安装到cuda的虚拟环境里面了。 现在就差这个库怎么安装到cuda的虚拟环境里了。 我尝试修改了代码: ``` elif platform.system() == 'Windows': cuda_include = 'C:\\Anaconda3\\envs\\onnx\\include' if platform.machine().endswith('64'): cuda_lib = 'C:\\Anaconda3\\envs\\onnx\\Lib\\x64' else: cuda_lib = 'C:\\Anaconda3\\envs\\onnx\\Lib\\Win32' extension_nvjpeg = Extension('nvjpeg', ['nvjpeg-python.cpp', 'src\\x86\\JpegCoder.cpp'], ['include', numpy.get_include(), cuda_include], [('JPEGCODER_ARCH',...

there is a error when i try to use this decodeing function in paddlepaddle dataloder, i guess maybe this package dosn't support multi-threading. -------------------------------------- C++ Traceback (most recent call last):...

save some image. ```python img = cv2.imread(img_file, cv2.IMREAD_COLOR) nj_jpg = nj.encode(img, 100) fp = open(output_img_path, "wb") fp.write(nj_jpg) fp.close() ``` if use cv2.imwrite only use 14% CPU, but nvjpeg use 26%...

Hi, I'm currently trying to install pynvjpeg on ubuntu 16.04 with pip install pynvjpeg. However, I face the awkward errors, the error logs are as follows: Building wheel for pynvjpeg...