Image classification viz

[reiffd7]

Description

This PR implements a new Classification Label Visualization block that provides flexible visualization options for classification predictions. The visualization supports both single-label and multi-label classification tasks with customizable display options.

Key features:

• Support for both single-label and multi-label classification tasks
• Customizable label positioning (9 positions: top, center, bottom × left, center, right)
• Flexible text display options (class name, confidence score, or both)
• Configurable styling (colors, text scale, padding, border radius)
• Automatic validation of prediction formats
• Efficient label spacing and positioning calculations

Type of change

• New feature (non-breaking change which adds functionality)
• This change requires a documentation update

How has this change been tested?

The visualization block has been tested with:

1. Single-label classification predictions
2. Multi-label classification predictions
3. Different text positioning options
4. Various styling configurations
5. Edge cases (no predictions, invalid formats)

Deployment considerations

• No additional dependencies required beyond existing supervision package
• Compatible with execution engine versions >=1.2.0,<2.0.0
• Low computational overhead as it caches annotation objects

Docs

• Docs updated with:
  • Comprehensive block description in LONG_DESCRIPTION
  • Parameter descriptions and examples in manifest
  • Usage examples for both single-label and multi-label scenarios
  • Clear explanation of text positioning options
  • Documentation of all customization parameters

[PawelPeczek-Roboflow]

Please acknowledge the changes that #761 introduced and adjust (read migration guide)

[PawelPeczek-Roboflow]

would be good if you could prepare examples of visualization

[PawelPeczek-Roboflow]

please resolve comments end of day for you, otherwise is not guaranteed to land in release

[reiffd7]

would be good if you could prepare examples of visualization

Images or Video?

[PawelPeczek-Roboflow]

image is enough