Jetson Nano + Cuda Image Compression, C ++, OpenCV

First level - You need to write a C ++ program that compresses images into JPEG format. A simple console application that does not accept parameters. Using OpenCV, it creates an RGB (BGR) image of 1920x1080 size, draws a rectangle on it and, most importantly, GPU compresses this image into JPEG format, while reducing the size to 1024x768, the JPEG compression level is 70. The resulting file is saved by the C ++ program in file result.jpg. Saving is done via std :: of-stream (i.e., you need to ensure that there is access to compressed data in the device’s memory) The second level is to modify the first program so that it receives images from the USB web-camera and saves them in JPEG, while the file name should correspond to the timestamp frame. On each frame through OpenCV draw a small red rectangle (i.e. guarantee access to an uncom-pressed image), and use CUDA tools to resize and compress in JPEG. The third level is to provide compression in H.264 with Jetson hardware (not Cuda, dedicated chip). Also receive a picture from a usb camera, draw a rectangle, compress it into an H.264 stream. Most importantly, a C ++ program must access compressed H.264 frames and inde-pendently write them to an .h264 file (frames are simply written sequentially to a single file without any special headers) via std :: ofstream. You can use GStreamer through Opencv :: VideoWriter if you can access h264 compressed frames If necessary, we will provide access to Jetson Nano via SSH. The result should be in the form of a .[login to view URL] archive with source codes and [login to view URL], assembly should be through CMake. The archive should contain assembly instructions, which also indicate additional dependencies that had to be installed for work.