1,555 images featuring text across three primary orientations including horizontal, multi-oriented, and curved layouts. The collection provides specialized data for scene text detection and recognition tasks involving complex geometric arrangements.
Use Cases
- Train scene text detection models to identify Curved text paths within natural images
- Benchmark OCR systems against Multi-Oriented and Horizontal text orientations
- Develop localization algorithms for arbitrary-shaped text regions using the orientation labels
Strengths
- 1,555 images of natural scenes containing text
- Categorizes text into three distinct orientations: Horizontal, Multi-Oriented, and Curved
- Specifically targets non-linear text paths and complex spatial layouts