Image Tagging

Build you first Image Tagging Project in just a few clicks with AI Studio. Here is how.

Sign up/ Login

Click here (hyperlink) if you need to set up a new account. Click here (hyperlink) if you need to login into your existing account.

(Insert Image)

Create a project

Step 1: Click on “New Project”.

(Insert Image)

Step 2: Choose Image Tagging as your Project Type.

(Insert Image)

Step 3: Click on Start from Scratch.

(Insert Image)

Step 4: Project Name and Description

Name your project and fill in the description (optional). It helps to define the specific problem statement in the description for clarity.

(Insert Image)

Step 5: Add labels.

List out all the labels you wish to train the model on. You can add more labels at a later stage too if you wish to update your model.

(Insert Image)

Step 6: Choose Categories.

Choose only those fashion object categories that you require the model to focus on.

(Insert Image)

Step 7: Click on Create Project.

(Insert Image)

You have just created your first Image Tagging Project on AI Studio in 7 simple steps and within few minutes.

Add datasets

After creating the project, a dialog box will appear for adding your datasets. For each of the labels, add a minimum of 100 images.

(Insert Image)

Train a model

Click on Start Processing. The AI Studio AutoML platform will pre-process your data and train your fashion AI model for image tagging based on the defined labels.

(Insert Image)

Evaluate model

Click on Analysis to see the evaluation metrics for the trained data.

(Insert Image)

You can view the pre-processing results by clicking on View Results and View Failed Images. The View Failed Images tells you why images failed. You can improve your model by uploading more data.

(Insert Image for View Images)

(Insert Image for View Failed Images)

Analyze results

Leverage the accuracy per label histogram, Confusion Matrix, Violin Plot, TSNE plot and PR Score to improve your dataset and enhance the accuracy rate of your model.

(Insert Image)

Deploy trained model

Step 1: Click on the Model tab

(Insert Image)

Step 2: Find List of Models

Under All Models at the bottom of the page, you will find the list of models built in the project.

(Insert Image)

Step 3: Choose Model

Choose the model you wish to deploy and click on View Info

(Insert Image)

Step 4: Test Model

Test you model using a new dataset it has not been exposed to as yet. Check for errors, accuracy and performance. Debug and improve the model if necessary.

(Insert Image)

Step 5: URL Deployment

After testing, click on Copy URL under the Deployment URL.

(Insert Image)

Step 6: Deploy Model

Deploy your trained model into your production workflow.

(Insert Image)