Russian School Notebooks with Handwritten Text and Detection Annotations

Images of school notebooks contain handwritten notes in Russian. The dataset provides end-to-end markup for training detection and OCR models. It was created by ai-forever and last updated in February 2023.

Use Cases

• Train an object detection model to locate text regions using bounding box annotations in COCO format.
• Develop an end-to-end OCR model to transcribe Russian handwritten text from the notebook page images.
• Fine-tune a text detection model on the specific layout and handwriting styles found in school notebooks.
• Benchmark segmentation or detection algorithms using the provided per-image annotations and category labels.

Strengths

• Annotations are provided in the standardized COCO format for object detection tasks.
• Includes end-to-end markup supporting both detection and text recognition model training.

Limitations

• The specific number of images, rows, and dataset size is unknown.
• Limited to Russian language handwriting, which may not generalize to other scripts or languages.
• Recency is unclear as the last known update was in early 2023.

Provenance

Source

ai-forever

Collection Method

Collection of images from school notebooks with handwritten notes, annotated for detection and OCR.

Freshness

Last updated 2023-02-09.