Shulou(Shulou.com)06/01 Report--

How to use ModelArts to achieve sign language recognition, in view of this problem, this article introduces the corresponding analysis and solution in detail, hoping to help more partners who want to solve this problem to find a more simple and feasible method.

According to statistics, more than 20 million people in China have hearing and language disorders, and every year, about 30, 000 babies are born without the opportunity to listen. Maybe it's because of the lack of ways to communicate with people who are not hard of hearing. If you look around, you may rarely see them, but they are real. Just as we are eager to communicate with others, they must also want to communicate with us, but there is a lack of a bridge between us. Now, with ModelArts, we can easily achieve this wish.

The steps for using ModelArts are not complicated, and mainly include:

1. Create Huawei Cloud account

two。 Apply for access key-used to authorize each application

3. Create an obs bucket-to store data

4. Authorized ModelArts

5. Import dataset

6. Automatic learning

7. Call online service prediction

Step 1: create Huawei Cloud account

Visit Huawei Cloud official website (https://www.huaweicloud.com/) and click the "Registration" button in the upper right corner of the website to enter the registration interface.

After entering the mobile phone number, SMS verification code and password as prompted on the web page, read and agree to the user agreement, click "agree Agreement" and register to complete the creation of Huawei Cloud (China) account.

After registration, please verify your real name, otherwise you will not be able to use ModelArts.

Step 2. Apply for an access key

Log in to Huawei Cloud using the registered account, hover your mouse over the user name in the upper right corner of the page, and click "my credentials" on the secondary menu to enter the AK key management interface.

Click * * access key * * on the left menu, and then click * * New access key * * to apply for SMS verification.

Enter the SMS verification code received, and a new access key will be created, and the browser will be prompted to download a file called "credentials.csv" containing the access key [Access Key Id] and private access key [Secret Access Key]. After applying for the access key at this time, please keep it properly.

Step 3. Creation of OBS bucket

Find "object Storage Service OBS" in the list of Huawei cloud services or directly access the OBS console (https://storage.huaweicloud.com/obs/).

Enter the OBS bucket management interface shown in the figure.

Click the right red button [create Bucket] to enter the creation interface and create a new OBS bucket.

Enter the name of the bucket and click "create now" to complete the creation of the obs bucket.

For more convenient data management, you can visit (https://developer.huaweicloud.com/tools) and download [OBS Browser+ tool] in [Business tools] below.

After downloading, use [AK login], and enter the access key [Access Key Id] and private access key [Secret Access Key] you just applied for in the input box to log in to the OBS bucket.

Step 4. Authorize ModelArts

Find [ModelArts] in the list of Huawei cloud services or visit the OBS console (https://console.huaweicloud.com/modelarts/) directly.

Click "Global configuration" at the bottom of the menu on the right, click "access Authorization", and select "use access key" for authorization.

Enter the access key [Access Key Id] and private access key [Secret Access Key] you just applied for in the input box, read and agree to the service statement, and click "agree to Authorization" to complete the ModelArts service authorization.

Step 5. Import the dataset

Open [OBS Browser+], open the previously created bucket, click "upload", click "Select folder", select the folder [train] in the provided compressed package, and click upload to upload it to the OBS bucket.

Open the browser, enter the [ModelArts console], click "data Management"-"dataset" in the menu on the right, and click "create dataset" on the right to enter the dataset creation page.

On the creation page, [dataset input location] and [dataset output location] select [New folder] and name them [input] and [output], respectively.

Finally, click "create" to complete the creation of the dataset.

Go back to the dataset management interface, select "more"-"Import" in the new dataset, select the [train] file previously uploaded to the OBS bucket, and click "OK" to import it into the dataset.

After the system completes the import, click "publish" in the dataset operation to complete the dataset creation and import.

Step 6. Automatic learning

Click "automatic Learning" on the left menu, select "Image Classification", select "existing dataset" from the dataset source, and select the dataset you just imported from the drop-down menu.

Then click "create Project", click "start training" on the right, click "next", click "submit", and the training task will start automatically. We just need to wait for the training to be finished.

Step 7. Call online service prediction

After the training, we can find the trained model in the menu [Model Management]-[Model] on the left.

Click the name of the model to view the characteristics of the model.

Click * * deploy * *-> * * online Services * * in the upper right corner of the interface to enter the online service deployment interface.

Select a free computing node (since a free computing node has been created, the paid node is shown in the example), click next, and click "submit" to start online service deployment.

When the online service is deployed, you can find the running online service in "deploy online"-- "online Service" on the left menu, and click "Forecast" to enter the prediction interface.

Click "upload", select the sign language image you want to recognize, and you can use the trained model for recognition.

This is the answer to the question about how to use ModelArts to achieve sign language recognition. I hope the above content can be of some help to you. If you still have a lot of doubts to be solved, you can follow the industry information channel for more related knowledge.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.