Difference between revisions of "Edge AI SDK/User Guide 300"

From ESS-WIKI
Jump to: navigation, search
Line 134: Line 134:
 
   
 
   
 
   
 
   
−
| colspan="3" rowspan="4" style="text-align: center; width: 122px;" |  
+
| colspan="1" rowspan="4" style="text-align: center; width: 122px;" |  
 
   
 
   
  

Revision as of 08:26, 4 June 2024

Quick Guide

Edge AI SDK caters to users who are comfortable with the machine learning (ML) experience. It can quickly evaluate computing performance for the CPU or GPU, and provides runtime results inferenced with ML.

Dduble click the icon ( in the red box ) on the Destop to start the Edge AI SDK App.

Step1.png

Quick Guide Video

How to Quickly Start AI Inferences

Step 1. Go to the “Quick Start (Vision)” page, as shown below. 

EAS-UserGuide-Vision-1.png

 

Step 2. Choose one application you want to activate and then confirm inference info, as shown below.

( 1. Object detection, 2. Face detection, 3. Person detection, 4. Pose estimation ) 

EAS-UserGuide-Vision-2.png

 

Step 3. Choose your video source which could be video clips or the USB Camera, as shown below.

Video ( Note: .MP4 only )

EAS-UserGuide-Vision-3.png  

or USB Camera 

EAS-UserGuide-Vision-4.png

 

Step 4. Choose one of the acceleration chipsets to execute selected inference application, as shown below.

( Note:  The available accelerator chipsets depend on your AI system and accelerator card. )

EAS-UserGuide-Vision-5.png  

Step 5. Another window will pop up to show the AI inference, as shown below.

Objection Detect.png

Object Detection

Object Detection is a computer vision task in which the goal is to detect and locate objects of interest in an image or video. The task involves identifying the position and boundaries of objects in an image, and classifying the objects into different categories. It forms a crucial part of vision recognition, alongside image classification and retrieval.

 

Objection Detect.png

Person Detection

Person Detection is based on object detection systems that can "detect human classification," i.e., have the data and training to classify the detected object as human. Person detection is used in many different sectors. These can be listed as security, insurance, nursing, health, and production. Technologies offer opportunities that can significantly increase customer satisfaction.

Person Detect.png

Face Detection

Facial Detection is the task of making a positive identification of a face in a photo or video image against a pre-existing database of faces. It begins with detection - distinguishing human faces from other objects in the image - and then works on identification of those detected faces.

Face Detect.png

Pose Estimation

Pose Estimation is a computer vision task where the goal is to detect the position and orientation of a person or an object. Usually, this is done by predicting the location of specific keypoints like hands, head, elbows, etc. in case of Human Pose Estimation.

Pose Detect.png

How to Quickly Start GenAI

Step 1. Go to the “Quick Start (GenAI)” page, as shown below.

EAS-UserGuide-GenAI-1.png

  Step 2. Choose application (chatbot) , as shown below.

EAS-UserGuide-GenAI-2.png

  Step 3. Choose one of the acceleration chipsets to execute selected chatbot application, as shown below.

EAS-UserGuide-GenAI-4.png

  Step 4. Another window will pop up to show the chatbot , as shown below.

EAS-UserGuide-GenAI-5.png

 

Monitor AI System Performance

Go to the “System Monitoring” page. This page shows payload and temperature for each chipset, as shown below. 

System-info-1.png

 

Evaluate the Benchmark Performance

Go to the “System Monitoring” page and select Visoin. Click the “Run” button to evaluate each chipset performance and wait for a few seconds to get FPS results, as shown below.

Benchmark-vison.png

  Go to the “System Monitoring” page and select GenAI. Click the “Run” button to evaluate each chipset performance and wait for a few seconds to get tokens results, as shown below.

Benchmark-genai.png

 



Install Information for Device

Edge AI System / Accelerator / OS AI Packages & Tools Note & Dependency


* ARK-3534 (Intel 13th Gen Core i9, RAM:64G) / EAI-3100 / Ubuntu 22.04
  • OpenVINO 2023.3.0-LTS (runtime)
  • OpenCV 4.7.0
  • Intel OpenVINO's benchmark_app
  • Vision Demos:
    • Object Detection 
    • Person Detection
    • Face Recognition
    • Pose Estimation
  • GenAI Demos:
    • Chatbot(Llama-2-7b) 


  • Ubuntu 22.04 intel-iotg
    • kernel:5.15.0-1027-intel-iotg
* AIR-150 (Intel 13th Gen Core i5, RAM:64G) / Hailo-8 / Ubuntu 22.04
  • OpenVINO 2023.3.0-LTS (runtime)
  • OpenCV 4.7.0
  • Intel OpenVINO's benchmark_app
  • Hailo AI Suite 1.0.2
  • Vision Demos:
    • Object Detection 
    • Person Detection
    • Face Recognition
    • Pose Estimation
  • GenAI Demos:
    • Chatbot(Llama-2-7b) 


  • Ubuntu 22.04 intel-iotg
    • kernel:5.15.0-1027-intel-iotg
* AIR-510 (Intel 12th Gen Core i9, RAM:64G) / RTX-A5000 / Ubuntu 22.04
  • OpenVINO 2023.3.0-LTS (runtime)
  • OpenCV 4.7.0
  • Deepstream 6.4
  • Intel OpenVINO's benchmark_app
  • Vision Demos:
    • Object Detection 
    • Person Detection
    • Face Recognition
    • Pose Estimation
  • GenAI Demos:
    • Chatbot(Llama-2-7b) 


  • Ubuntu 22.04 intel-iotg
    • kernel:5.15.0-1027-intel-iotg
* AIR-030 (Jetson AGX Orin RAM:32G) / None / Ubuntu 20.04
  • Deepstream 6.2
  • jetson_benchmarks
  • Vision Demos:
    • Object Detection 
    • Person Detection
    • Face Recognition
    • Pose Estimation
  • GenAI Demos:
    • Chatbot(Llama-2-7b) 


  • Ubuntu 20.04
    • kernel:5.10.104-tegra
  • JetPack 5.1.2
  • L4T: 35.3.1












Installation

NOTE:

* Not Using (sudo su or sudo ) to install

* An active Internet connection is required.

* Don't "apt upgrade" before installation 

$tar zxvf Edge_AI_SDK-installer-<version>.tar.gz
$./Edge_AI_SDK-installer.run

EAS-installer-run-s1.png

  • License Aggreement 
EAS Install s2.png
  • System Information ( by target platform ) 
Installer-3.0.0-p1.png
  • Select AI Accelerator SDK 

Installer-3.0.0-p2.png

  • Select GenAI SDK 

Installer-3.0.0-p3.png

 

Activation

  • User Information. Please fill in the required information one by one, especially a correct email. The following activation will need an available email. Then press the Next button to proceed. 

EAS-Activataion-S1.png


 

  • License agreement. Please review EULA and then press the Activation button if the content is agreed.

EAS-Activataion-S2.png


* Activation Code: If it’s available to access internet, a dialog will pop up to ask you to enter a token for activation.

EAS-Activataion-S3.png


 

Uninstall

cd /opt/Advantech/EdgeAISuite

./unInstall.sh

 

 

 

Retrieved from "https://ess-wiki.advantech.com.tw/wiki/index.php?title=Edge_AI_SDK/User_Guide_300&oldid=35593"