765 views|7 replies

224

Posts

0

Resources
The OP
 

【STM32MP135F-DK】5- Evaluation on artificial intelligence - object recognition [Copy link]

This post was last edited by qiao--- on 2023-12-11 00:44

After having a basic understanding of openSTLinux, we found that this system can run some artificial intelligence applications. In this evaluation, we will evaluate the performance of this board in artificial intelligence. X-LINUX-AI is a free open source software package dedicated to AI. It is a complete ecosystem that allows developers using OpenSTLinux to create AI-based applications very easily. The following is the ecosystem of openSTLinux.

The following is the overall framework of X-LINUX-AI

Specifically, X-LINUX-AI is a software package that allows developers to easily run some AI applications on their systems, so we need to build the environment for this AI software package.

We first connect our development board to the router with an Ethernet cable and configure it with an IP address in the same network segment as the router, so that our development board can access the Internet. For information on setting up the network environment, you can read my third review.

We can ping Baidu to see if our environment is set up. After the network environment is set up, we can download the software package.

First execute the following command in the console to configure the AI OpenSTLinux package repository

cd /var/cache/apt/archives
wget http://extra.packages.openstlinux.st.com/AI/5.0/pool/config/a/apt-openstlinux-ai/apt-openstlinux-ai_1.0_armhf.deb
apt-get install ./apt-openstlinux-ai_1.0_armhf.deb
 apt-get update

Then install all the X-LINUX-AI packages, I have already installed them here

apt-get install packagegroup-x-linux-ai

Restart the demo launcher to update the system interface

 systemctl restart weston-graphical-session.service

At this point the AI environment has been installed.

We enter the following command in the terminal to see the version of the AI environment

x-linux-ai -v

At this point, our interface has several more AI examples, as shown below:

When we click the two arrows, the interface will jump to the next page. There are two examples of photo classification and object recognition. Let's install the camera and run the object recognition example.

1


It feels like the accuracy is not very high, and the frame rate shown next to it is not very high either.

However, for the 1GHZ mp135, it is still quite capable of running artificial intelligence applications, after all, it is an entry-level SOC chip.

Summary: stm32mp135 can run AI-related applications, but its performance is limited, and its accuracy and recognition speed are poor.

This post is from stm32/stm8

Latest reply

"When we click on the two arrows, the interface will jump to the next page. There are two examples of photo classification and object recognition." --After the installation, several new icons appeared, for target detection. I don’t know what each one is used for and what is the difference between them?   Details Published on 2024-7-3 17:13
 

6010

Posts

6

Resources
2
 

Does this require some learning? Nowadays, small supermarkets use image recognition for cashier weighing.

This post is from stm32/stm8

Comments

Yes, you still have to train the model yourself. I'll give you a lesson when I have time.  Details Published on 2023-12-11 11:01
 
Personal signature

在爱好的道路上不断前进,在生活的迷雾中播撒光引

 

623

Posts

0

Resources
3
 

This speed is fast enough. 15fps is completely suitable for civilian use - this is how I understand it

This post is from stm32/stm8

Comments

I always feel that the screen is stuck when recognizing, especially when I turn the direction  Details Published on 2023-12-11 10:59
 
 

2865

Posts

4

Resources
4
 

Is this recognition offline or online? If it is online, the scope of application is not that large.

This post is from stm32/stm8

Comments

Offline, online related programs are not written, I don't know how to do it yet, I still need to study  Details Published on 2023-12-11 10:58
 
 
 

224

Posts

0

Resources
5
 
bigbat posted on 2023-12-11 10:53 Is this recognition offline or online? If it is online, the scope of application is not that large.

Offline, online related programs are not written yet, I still don't know how to do it for the time being, I still need to study


This post is from stm32/stm8
 
 
 

224

Posts

0

Resources
6
 
jobszheng5 posted on 2023-12-11 09:14 This speed is fast enough. 15fps is completely suitable for civilian use - this is how I understand it

I always feel that the screen is stuck when recognizing, especially when I turn the direction


This post is from stm32/stm8
 
 
 

224

Posts

0

Resources
7
 
Qintianqintian0303 posted on 2023-12-11 09:01 Does this require some learning? Nowadays, small supermarkets use image recognition for cashier weighing

Yes, you still have to train the model yourself. I'll give you a lesson when I have time.


This post is from stm32/stm8
 
 
 

21

Posts

0

Resources
8
 

"When we click on the two arrows, the interface will jump to the next page. There are two examples of photo classification and object recognition."

--After the installation, several new icons appeared, for target detection. I don’t know what each one is used for and what is the difference between them?

This post is from stm32/stm8
 
 
 

Just looking around
Find a datasheet?

EEWorld Datasheet Technical Support

Copyright © 2005-2024 EEWORLD.com.cn, Inc. All rights reserved 京B2-20211791 京ICP备10001474号-1 电信业务审批[2006]字第258号函 京公网安备 11010802033920号
快速回复 返回顶部 Return list