Ultralytics + OpenVINO 2 | Intel Software

[00:00:03]The ease of use and simplicity of ultralytics  plus the power of OpenVINO, in 2 minutes Ultralytics can help you do many things if  you are building vision based applications​ Check out their website ​ Let me show you how to run inference using OpenVINO open a virtual environment, get there​ pip install ultralytics​ open a new python file,​ Import YOLO from ultralytics​ now we need a model,​ go to docs.ultralytics.com​ let's pick a YOLO 26 segmentation model,​ Line #3 will bring the model over, ​ and this line will convert the model to  OpenVINO format, even quantized to INT8​ and there are many more options..​ Line 9 will run the model on and image, and will use the intel NPU, you can  change the device to CPU GPU or even AUTO​ run the script ​ and the segmentation is working​ if you are looking for the model itself ​ it will be stored in a dedicated directory, bin file, xml file and a file for the meta data​ all kind of information here, all the  labels, data-set used and more, pretty handy Once you have the model, I want to show you how ultralytics API  opens up so many easy ways to use it there are multiple ways to perform inference, ​ let's go to the ultralytics documentation ​ to predict​ and you can see here all the input formats to model-predict​ you can of course use an image, a URL, a video and much more​ for each type you have the code snippet to use.​ Let me copy the url of this youtube video​ we'll use that URL as the source​ and we'll use the model to predict  from this source, enable streaming, show the video and even run  it on our integrated GPU​ run the script​ and it is running, ​ This is faster than the normal speed, very nice​ the truth is that I don't even need a script,​ I can run a command line, yolo, predict, the  path to the model directory, path to the source, and yes please show the video​ one line, and it is running, nice.

[00:02:41]Ultralitics + OpenVINO is a winning combination.​ The simplicity and robustness of Ultralytics API ​ With the inference speed more advantages of OpenVINO​ץ Check it out ​