Introduction
Have you ever heard of Intel® Developer Cloud (DevCloud)? DevCloud is a cloud environment provided by Intel that allows you to perform evaluations from anywhere as long as you create an account for free!
You can use not only server-class CPUs, but also CPUs for embedded environments used in industrial environments, so you can do simple evaluations without having to buy a device!
In this article, we will show you how to access DevCloud and how to perform object detection using OpenVINO™, which is provided as a sample.
(Note: This article is based on information as of February 2024.
Three types of DevCloud environments
Here is the Intel ® Developer Cloud homepage
https://www.intel.com/content/www/us/en/developer/tools/devcloud/overview.html
There are three environments available on the site. I will briefly explain each of them. Basically, you can access any of the environments by creating an account, which will be explained later.
- Accelerate AI and HPC: You can use the latest generation of Intel® Xeon® scalable processors and Intel® Iris® Xe MAX GPU series, Intel's server GPUs.
- Optimize for the Edge: Use the latest generation as well as previous generation Intel® Core® processors
- Build Multiarchitecture and FPGA Applications: Basically an environment for evaluating oneAPI, using Intel® Xeon® scalable processors as well as Intel® Iris® Xe MAX and Intel® Arria® 10 FPGAs and Intel® Stratix® 10 FPGAs. Stratix® 10 FPGAs, as well as Intel® Iris® Xe MAX, Intel® Arria® 10 FPGAs, and Intel® Stratix® 10 FPGAs.
In this article, we will use the "Optimize for the Edge" environment.
Create a DevCloud account
From the Overview page Optimize for the Edge" page from the Overview page
https://www.intel.com/content/www/us/en/developer/tools/devcloud/edge/overview.html
First, click Create an Intel Developer Cloud Account
Then click Create an Account to create a new account
Enter your email address, name, language, country, and password.
You will receive an email confirmation to the email address you entered. Enter the code from the email into the Verification code field and click on Create an account .
After a page of information such as your purpose, affiliation, and the newsletter you wish to subscribe to, click Submit and your account will be created. Click Submit and your account is created. After creation, you will receive an email notification that your account has been created, and you will also see on your screen that your Intel® Developer Cloud account has been created.
Your account creation is now complete!
Next, click on the "Optimize for the Edge" environment
Intel® Developer Cloud for the Edge
The Optimize for the Edge environment is actually called Intel® Developer Cloud for the Edge. Let's use this environment to run OpenVINO™!
https://www.intel.com/content/www/us/en/developer/tools/devcloud/edge/overview.html
This is the same page as when you created your account, but click on Get Started under Bare Metal Development in the middle of the page.
After that, log in with the account you just created.
After logging in, setup of your cloud environment will begin.
After a short wait, you will see a console screen like this
The environment is prepared in the form of a Jupyter Note Book, and each sample can be run step by step.
Let's run Accelerated Object Detection , the object detection sample on the far right.
Accelerated Object Detection Demo
Running the sample opens the Jupyter Notebook, which provides a description of the sample.
The execution itself is done by clicking on the symbol at the top, and each step is executed.
Some parts may take a while to execute, so you will need to check the output as you go along.
For example, downloading the model in [3] and converting the model in mo in [4] will take some time, so you will need to wait a bit.
In the case of [3], you should proceed after the message "All files that were downloaded!
After proceeding through the steps, you will finally come to the " Run the demo " step.
After running the demo, you will be able to select the CPU and GPU you wish to run the demo on.
For example, if you want to run on the Intel®Iris Xe graphics GPU built into the Intel® Core™ i7-1185G7E, select the following and click Submit
After you submit, the job will be queued.
When the Job is ready to run, it will run automatically.
After Pre Infer Progress, Inference Progress, and Post Infer Progress are all Done and a little while has passed, Display output, which had been dimmed, can be selected and Plot results will be displayed.
Click "Display Outout" to see a video of the inference results.
If you click Telemetry, you can see the results of the runtime in a separate tab and check the power, memory usage, CPU usage, etc.
Plot results will show you the Inference Time and Frame per second in a graph.
Once you return to Run the Demo, select a different target node or target architecture and submit the same way, you can compare the execution results in Plot results.
For example, let's try processing on a CPU instead of a GPU, even though the CPU is the same this time.
Also, Pre Infer Progress, Inference Progress, and Post Infer Progress are all set to Done, and you can compare the results with the previous one by clicking "Plot results".
Conclusion
This time, we used the Intel® Developer Cloud to perform Object Detection with OpenVINO™.
This environment also contains a variety of other samples, so it is useful for those who are new to OpenVINO™, as well as for those who want to see what OpenVINO™ can do and estimate devices in an edge environment.
In addition, there is also an environment for running generative AI, so please take a look at what OpenVINO™ can do for you!