Modelarts image classification and object detection

Catalog

1 establish obs Barrel in two ways ：

1.1 download obsBrowers+

1.2. In the console of Huawei cloud official website obs Create buckets in the management service ​ edit

2. Go to obs Import data set into bucket

2.1 Dataset reference website : 

2.2 step  

3. Automatic learning

3.1 Image classification - Mask detection

3.2 Object detection - flowers

---

1 establish obs Barrel in two ways ：

Are all the same , Any one will do .

1.1 download obsBrowers+

In my voucher in the user list of Huawei cloud official website , New access key , Get the download file , Enter the corresponding information to log in

 

1.2. In the console of Huawei cloud official website obs Create buckets in the management service

2. Go to obs Import data set into bucket

Here are - adopt obsBrowser+ Import

2.1 Dataset reference website : 

AI Gallery_ data _ developer _ Hua Wei Yun (huaweicloud.com)

2.2 step  

  Upload after creating a new folder

3. Automatic learning

3.1 Image classification - Mask detection

Here is the data set of masks

First create the project

Fill in information , Import obs Data in barrel , Here simply create a  

  After synchronizing the data , Start tagging , Start model training after marking

Training done , It's free here , It has training effect  

Then deploy and go online , It's free here , After the launch, you can test  

Here is the test , Mark out the probability of being human ...

3.2 Object detection - flowers

The data set used here is Huawei aigallery Data set of flowers

The process of creating a project is saved ... Relatively simple

Label datasets , Start training

The training settings are as follows ： It's paid , Remember to get coupons , Otherwise, I'll owe you ...

Then submit and start training ...

 

 

Because the training time is relatively short , So it's not very accurate ... 

Then it can be deployed online for testing ： 

The test results are relatively good , It can be seen that it is dandelion , Because it uses the pictures in the data set