An AI model uses a specific set of visual elements called 'expressive shapes'. These shapes are chosen based on quality ratings and size. A consistent method is used to match these shapes to text.
By Pawan Nayar · from the LLOS AI Learning record
An AI model uses a specific set of visual elements called 'expressive shapes'.
These shapes are chosen based on quality ratings and size. A consistent method is used to match these shapes to text.
The AI model uses a special collection of visual shapes. These shapes are stored in two main files. One file lists all shapes with details like name, word, and category. The other file has quality ratings for each shape, from 0 to 5 stars. Only shapes with a 5-star rating are used. Also, the shape's image data must be large enough. This ensures the shape looks good when displayed. This filtering process creates a smaller, high-quality set of shapes. To match a shape to text, the AI model looks at keywords. Each shape has a 'keyword blob' which combines its word, context, category, and tags. The model compares keywords from the text to these keyword blobs. The shape with the most matching keywords is chosen.
A developer is building a new feature for an AI model. This feature needs to show a visual shape next to each generated sentence. The developer uses the defined process: first, they filter the master list of shapes to include only 5-star rated shapes with large enough SVG data. Then, for each sentence, the developer extracts keywords. The developer compares these keywords to the 'keyword blob' of each filtered shape. The shape that best matches the sentence's keywords is then displayed. If the developer does not follow this, low-quality or irrelevant shapes might appear.
Resolved from this lesson's own record — real definitions, not invented for this page.
More from the same record, pulled by shared topic — not written new for this page.
A large collection of expressive shapes exists. This collection is the main source.
The system uses 50 hand-crafted noun silhouettes. These are organized into five domains.
A main library holds many shapes. A separate library stores special shapes, like religious symbols.
A project has a special collection of approved visual elements. This collection includes many emojis and shapes.
An AI model uses a specific set of visual elements called 'expressive shapes'. These shapes are chosen based on quality ratings and size. A consistent method is used to match these shapes to text.