Huawei AI Certification Training

HCIA-AI
ModelArts Lab Guide
ISSUE: 3.5

Huawei Technologies Co., Ltd.

1
Copyright © Huawei Technologies Co., Ltd. 2023. All rights reserved.
No part of this document may be reproduced or transmitted in any form or by
any means without prior written consent of Huawei Technologies Co., Ltd.

Trademarks and Permissions

and other Huawei trademarks are trademarks of Huawei Technologies

Co., Ltd.
All other trademarks and trade names mentioned in this document are the
property of their respective holders.

Notice
The purchased products, services and features are stipulated by the contract
made between Huawei and the customer. All or part of the products, services
and features described in this document may not be within the purchase
scope or the usage scope. Unless otherwise specified in the contract, all
statements, information, and recommendations in this document are
provided "AS IS" without warranties, guarantees or representations of any
kind, either express or implied.
The information in this document is subject to change without notice. Every
effort has been made in the preparation of this document to ensure accuracy
of the contents, but all statements, information, and recommendations in this
document do not constitute a warranty of any kind, express or implied.

Huawei Technologies Co., Ltd.

Address: Huawei Industrial Base Bantian, Longgang Shenzhen 518129
People's Republic of China
Website: https://e.huawei.com
 HCIA-Artificial Intelligence ModelArts Lab Guide Page 1

Huawei Certification System

Huawei Certification is an integral part of the company's "Platform +
Ecosystem" strategy, and it supports the ICT infrastructure featuring "Cloud-
Pipe-Device". It evolves to reflect the latest trends of ICT development. Huawei
Certification consists of three categories: ICT Infrastructure Certification, Basic
Software & Hardware Certification and Cloud Platform & Services Certification,
making it the most extensive technical certification program in the industry.
Huawei offers three levels of certification: Huawei Certified ICT Associate
(HCIA), Huawei Certified ICT Professional (HCIP), and Huawei Certified ICT
Expert (HCIE).
Huawei Certification covers all ICT fields and adapts to the industry trend of
ICT convergence. With its leading talent development system and certification
standards, it is committed to fostering new ICT talent in the digital era, and
building a sound ICT talent ecosystem.
HCIA-AI V3.5 certification is intended for cultivating and conducting
qualification of engineers who are capable of creatively designing and
developing AI products and solutions using machine learning and deep learning
algorithms.
An HCIA-AI V3.5 certificate proves that you:
 Have understood the development history of AI, Huawei Ascend AI
system, and Huawei full-stack AI strategy in all scenarios;
 Have mastered traditional machine learning and deep learning
 Are able to use the MindSpore framework to build, train, and deploy
neural networks;
 Are competent in sales, marketing, product manager, project
management, and technical support positions in the AI field.
HCIA-Artificial Intelligence ModelArts Lab Guide Page 2
 HCIA-Artificial Intelligence ModelArts Lab Guide Page 3

About This Document

Overview
This document is intended for trainees who are preparing for the HCIA-AI
certification examination or readers who want to achieve basic AI knowledge.
After studying this document, you will be able to understand and use Huawei
Cloud ModelArts ExeML.

Description
This document presents one exercise on Huawei Cloud ModelArts ExeML.
 Exercise 1: How to use Huawei Cloud ModelArts ExeML

Knowledge Required
This course is for Huawei's development certification. To better understand this
course, familiarize yourself with:
 Basic computer operations.

Lab Environment
 Accessible network
 HCIA-Artificial Intelligence ModelArts Lab Guide Page 4

Content

About This Document.......................................................................3

Overview..................................................................................................................................... 3
Description.................................................................................................................................. 3
Knowledge Required.................................................................................................................... 3
Lab Environment......................................................................................................................... 3
1 ExeML........................................................................................... 5
1.1 Introduction........................................................................................................................... 5
1.2 Objectives.............................................................................................................................. 5
1.3 Lab Environment................................................................................................................... 5
1.4 Flower Recognition Exercise.................................................................................................. 6
1.4.1 Creating a Project............................................................................................................... 7
1.4.2 Labeling Data..................................................................................................................... 8
1.4.3 Training a Model................................................................................................................. 9
1.4.4 Deploying a Service and Performing Prediction................................................................10
 HCIA-Artificial Intelligence ModelArts Lab Guide Page 5

1 ExeML

1.1 Introduction
ExeML, a service provided by ModelArts, automates model design, parameter
tuning and training, and model compression and deployment using labeled data.
ExeML is free of coding and does not require you to have experience in model
development, meaning that it is beginner friendly. The exercise covered in this
course will help you understand and use image classification, object detection,
and predictive analysis.
Image classification identifies a class of objects in images. An image
classification model can predict the label of an image, and applies to scenarios in
which image classes are distinguishable.

1.2 Objectives
The exercise in this course uses flower recognition as an example, to help you
quickly create and learn about image classification models.

1.3 Lab Environment

If you are using ModelArts for the first time, authorize ModelArts to access
Huawei Cloud Object Storage Service (OBS). Without doing so, you will not be
able to create jobs on ModelArts. The procedure is as follows.
 HCIA-Artificial Intelligence ModelArts Lab Guide Page 6

Figure 1-1 ModelArts console

1.4 Flower Recognition Exercise

The ModelArts ExeML page consists of two parts. The upper part lists the
supported project types. You can click Create Project to create an ExeML
project. The lower part displays the list of created ExeML projects. You can filter
projects by project type in the upper right corner of the list, or enter key words in
the text box and click to search for projects that contain those key words.

Figure 1-1 ExeML

The procedure of the exercise consists of four parts:
 Creating a project: To use ModelArts ExeML, first create an ExeML project.
 HCIA-Artificial Intelligence ModelArts Lab Guide Page 7

 Labeling data: Upload images and label them by class.

 Training a model: After data labeling, train a model.
 Deploying a service and performing prediction: Deploy the model as a
service and perform real-time prediction.

1.4.1 Creating a Project

Step 1 Set parameters.
On the ExeML page, click Create Project in Image Classification.
The Create Image Classification Project page is displayed.

Figure 1-1 Creating a project

Parameters:
Billing Mode: Pay-per-use by default
Name: Enter a project name.
Dataset Source: Select the OBS path where the dataset is stored.
You must create an empty OBS folder and select this folder as the training data
path. (Click the bucket name, and then click Create Folder.) Alternatively,
import the data to OBS in advance. In this example, click the following data link
and decompress the downloaded file package, and upload the data in the train
folder to the OBS/ExeML/flower/train directory.
For details about how to upload data, see
https://support.huaweicloud.com/intl/en-us/modelarts_faq/modelarts_05_0013.ht
ml.
Data link:
https://certification-data.obs.cn-north-4.myhuaweicloud.com/ENG/HCIA-AI/V3.5/
chapter5/ExeML.zip
 HCIA-Artificial Intelligence ModelArts Lab Guide Page 8

Step 2 Click Create Now.

Click Create Now. The ExeML project is then created.

1.4.2 Labeling Data

Step 1 Upload images.
After an ExeML project is created, the data labeling page is automatically
displayed. Images uploaded to the OBS directory configured during project
creation are automatically loaded. Labeled images are displayed in the Labeled
tab.

Figure 1-1 Data labeling for image classification

Notes:
 The images to be trained must be classified into at least two classes, and
each class must contain at least five images. That is, at least two labels are
available and the number of images for each label is not fewer than five. In
this example, the dataset contains five labels: tulips, daisy, sunflowers,
roses, and dandelion.
 You can add multiple labels to an image.

Step 2 Label the images.

Click Unlabeled and select the unlabeled images you want to label, or select
Select Images on Current Page in the upper right corner to select all images
on the page. Then, input a label or select an existing label from the drop-down
list, and then press Enter. Then, click OK.
 HCIA-Artificial Intelligence ModelArts Lab Guide Page 9

Figure 1-1 Image labeling for image classification

Step 3 Modify labels.
Click the Labeled tab, select an image, enter a new label name in the Label
text box on the right, and click OK.
You can select multiple images at a time and modify them in batches.

1.4.3 Training a Model

After labeling the images, train a model. Before training, set parameters. Images
to be trained must be classified into at least two classes, and each class must
contain at least five images. Therefore, before training, ensure that the labeled
images meet the requirements. Otherwise, the Train button remains
unavailable.

Step 1 Set parameters.

Use default settings or configure them.
 HCIA-Artificial Intelligence ModelArts Lab Guide Page 10

Figure 1-1 Training Configuration

Parameters:
max training time (minute): If the training is not completed within the
maximum training duration, the training is forcibly stopped. Enter a larger value
to avoid this situation.

Step 2 Train a model.

After setting, click Train. After the training, you can view the training result on
the Train Model tab page.

1.4.4 Deploying a Service and Performing Prediction

Step 1 Deploy a service.
After model training, deploy a model version with satisfactory accuracy and in
the Completed status as a service. To do so, click Deploy in Version Manager
of the Train Model tab page. After the deployment, choose Service
Deployment > Real-Time Services to view the deployed service.
 HCIA-Artificial Intelligence ModelArts Lab Guide Page 11

Figure 1-1 Deploy

Step 2 Test the service.
After the model is deployed, test the service using an image. The test data is
stored in OBS/flower/test/daisy.jpg. On the Deploy Service tab, click Upload
and upload an image. Then, click Predict. The test result is displayed in the right
pane. There are five labels for data labeling: tulip, daisy, sunflower, rose, and
dandelion. In the prediction, "daisy" gets the highest score, so the classification
result is "daisy".

Figure 1-1 Service test