ASRock Industrial AI Guru Software User Manual

ASRock Industrial AI Guru Software

Specifications

• Product Name : AI Guru
• Version: 1.0
• Published Date : March 22, 2024

Product Usage Instructions

Chapter 2: Installation

Operating System

Ensure your computer has the latest version of drivers by referring to ASRock Industrial’s NUC Ultra 100 BOX/NUCS Ultra 100 BOX Series webpage.

Environment Installation

1. Download AI Guru software from the provided link.
2. Extract the downloaded file.
3. Run Install env.bat for driver installation.
4. Follow on-screen instructions for Python setup.
5. Restart the computer after installation completion.

Executing AI Guru

1. Click on AI_Guru.exe to start execution.
2. Run inference with the available AI models.

Chapter 3: AI Guru Models

Face Recognition

1. Select Face Recognition as the AI model and your inference device.
2. Choose a specific human face image for inference input.
3. Select the folder to be searched for.
4. Adjust confidence level for accuracy configuration.
5. Apply settings and wait for inference results.

Object Detection

The Object Detection model identifies 80 objects from a camera device or video file. Refer to Chapter 4 Appendix for the object detection list.

FAQ

• Q: How long does the installation process take?
  • A : The installation process, including driver installation and setup, may take approximately 30 minutes to complete.

Chapter 1 Introduction

ASRock Industrial’s AI Guru software incorporates seven pre-trained AI models, powered by Intel® Core™ Ultra processors (Meteor Lake-H), leveraging the Intel® Distribution of OpenVINO™ Toolkit. Users can download the AI Guru software from ASRock Industrial’s website, and initiate AI inference on the NUC Ultra 100 BOX/NUCS Ultra 100 BOX Series.

In this manual, Chapter 1 briefly introduces AI Guru. Chapters 2 and 3 contain the step-by-step guide of the AI Guru software installation and demonstrations of seven AI model inferences. Chapter 4 includes BIOS configuration.

Because ASRock Industrial’s AI Guru software is based on pre-trained models from the Intel® Distribution of OpenVINO™ Toolkit, AI inference results may be limited and can be further fine-tuned.

If you require technical support related to the AI Guru, please visit our forum for further information.
https://forum.asrock.com/forum_topics.asp?FID=21&title=asrock-industrial-ai- guru

AI Guru Contents

ASRock Industrial’s AI Guru has the following seven pre-trained AI models powered by Intel®, with options of CPU, GPU and NPU to select for inference device.

1. Face Recognition
2. Object Detection
3. Human Pose Estimation
4. Image Inpainting
5. Face Detection
6. Gaze Estimation
7. Interactive Face Detection

Chapter 2 Installation

Operating System

To execute ASRock Industrial’s AI Guru software, it is recommended to operate under Microsoft® Windows® 11 64-bit operating system, and to use Meteor Lake (Intel® Core™ Ultra processors) to have the optimal experience.

Environment Installation

1.  Refer to ASRock Industrial’s NUC Ultra 100 BOX/NUCS Ultra 100 BOX Series webpage to ensure your computer have the latest version of drivers.
2. Restart the computer after both GPU and NPU drivers are installed, and then you will see GPU and NPU performance in your Task Manager.
3. Download ASRock Industrial’s AI Guru software
   • Global : https://download.asrock.com/IPC/Drivers/All/Others/AI_Guru(v1.0.0.1).zip
   • China: ftp://asrockchina.com.cn/Drivers/All/Others/AI_Guru(v1.0.0.1).zip
4. Extract the downloaded AI Guru compressed file.
5. Click on “Install env.bat” to proceed with the installation of relevant drivers and tools. The installation path and settings will apply default configurations.
6. When you see “Install Python” setup, select “Install Now” (recommended), or you may choose your customized location and features. Wait until the setup is successful.
7. When the “Self-extracting archive” box pops up, you may apply default path (recommended) or a customized path. Then click “Extract.
8. Wait until the execution is completed, press any key to continue. Then restart the computer to execute the AI Guru.

Executing AI Guru

1. Click on “AI_Guru.exe” to start execution.
2. You may start to run inference with seven AI models.
   • The three installations may take about 30 minutes to complete.

Chapter 3 AI Guru Models

Face Recognition

The Face Recognition model identifies a specific person’s face from a given face database.
Each image should contain frontally-oriented faces with decent quality.

1. Select “Face Recognition” as AI model and your inference device.
2. Select a specific human’s face image as the input for inference.
3. Choose the folder to be searched for.
4. Adjust your preferred confidence, from 0 to 100, as your accuracy configuration.
   • The demonstration selects seven images for reference input and 80% of accuracy rate. Seven to ten reference inputs and more than 80% of accuracy rate are recommended, but the actual inference results may vary depending on factors such as image quality.
5. Press “Apply” and wait for the inference. AI Guru starts to recognize the given face and displays the results in the “Output” folder
   • The pictures in the Output folder are only shortcuts and will be deleted every time a new AI inference starts.

Object Detection

The Object Detection model identifies 80 objects from either camera device or video file, and performs an independent analysis of each frame. See Chapter 4 Appendix for the object detection list.

1. Select “Object Detection” as AI model and your inference device.
2. Select a video or connect a camera device, and press “Apply.” Then wait for the AI inference results.
3. AI model starts to run the inference. You may see the FPS (frame per second, or average rate of video frame processing) and latency (average time required to process one frame) on the upper left. In our demonstration, a giraffe and zebras are individually detected at 97.3% and more than 50% of the correspondence rate.

Human Pose Estimation

From the body skeleton, it is estimated that someone is slowly moving forward.

The Human Pose Estimation model detects multi-person pose in an image/video/camera by body skeleton. A body skeleton consisting of keypoints and connections between them.

1. Select “Human Pose Estimation” as AI model and your inference device. NPU is not supported.
2. Select your input source (image/video/camera), and press “Apply.” Then wait for the AI inference results.
3. AI model starts to run the inference. In our demonstration, you may see key points connecting into a body skeleton, which can be used to estimate human poses.

Image Inpainting

The Image Inpainting model estimates suitable pixel information and fill holes in a tampered image, creating something similar to the original image.

1. Select “Image Inpainting” as AI model and your inference device.
2. Select an image as input source, and press “Apply.”
3. You may start to edit.
4. Press Space or Enter to paint.
   • Press Space or Enter to paint. Press Esc or Q to quit.

Face Detection

The Face Detection model displays the boundary and feature point of a human face.

1. Select “Face Detection” as AI model and your inference device.
2. Select an image, a video or connect a USB camera as the input source, and press “Apply.”
3. AI model detects human faces.

Gaze Estimation

The Gaze Estimation model performs inference on gaze estimation model by using auxiliary networks of face detection, head pose estimation, facial landmarks detection and open/closed eye state detection.

1. Select “Gaze Estimation” as AI model and your inference device.
2. Select an image, a video or connect a USB camera as the input source, and press “Apply.”
3. AI model detects multi-person gaze angles.

Interactive Face Detection

The Interactive Face Detection model performs up to five simultaneous inferences to identify faces, including age/gender, head pose, emotions, facial landmarks, and anti-spoof detection networks.

1. Select “Interactive Face Detection” as AI model and your inference device. NPU is not supported.
2. Select an image, a video or connect a USB camera as the input source, and press “Apply.” Wait for the AI inference results.

Chapter 4 How to Set Performance Mode

For a better experience, it is recommended to follow the steps below to set performance mode. Refer to Chapter 5 of ASRock Industrial’s NUC Ultra 100 BOX/NUCS Ultra 100 BOX Series user manual for detailed entering BIOS setup guide.

1. Enter Bios.
2. Select Advanced → CPU Configuration → press Enter.
3. Select CPU Operating Mode → press Enter.
4. Select Performance Mode → press Enter.
5. CPU Operating Mode changes into [Performance Mode].
6. Press Esc
7. Select Exit → Save Changes and Exit → press Enter.
8. Select Yes → press Enter.

Appendix

Object Detection List

Documents / Resources

| ASRock Industrial AI Guru Software [pdf] User Manual
AI Guru Software, AI Guru, Software
---|---

References

• ASRock Industrial AI Guru - ASRock Forums

Read User Manual Online (PDF format)

Download This Manual (PDF format)

Download this manual  >>