README.md

metadata

tags:
  - ocr
  - document-processing
  - deepseek
  - deepseek-ocr
  - markdown
  - uv-script
  - generated

Document OCR using DeepSeek-OCR

This dataset contains markdown-formatted OCR results from images in Alysonhower/test using DeepSeek-OCR.

Processing Details

• Source Dataset: Alysonhower/test
• Model: deepseek-ai/DeepSeek-OCR
• Number of Samples: 1
• Processing Time: 1.5 minutes
• Processing Date: 2025-10-23 13:06 UTC

Configuration

• Image Column: image
• Output Column: markdown
• Dataset Split: train
• Resolution Mode: gundam
• Base Size: 1024
• Image Size: 640
• Crop Mode: True

Model Information

DeepSeek-OCR is a state-of-the-art document OCR model that excels at:

• 📐 LaTeX equations - Mathematical formulas preserved in LaTeX format
• 📊 Tables - Extracted and formatted as HTML/markdown
• 📝 Document structure - Headers, lists, and formatting maintained
• 🖼️ Image grounding - Spatial layout and bounding box information
• 🔍 Complex layouts - Multi-column and hierarchical structures
• 🌍 Multilingual - Supports multiple languages

Resolution Modes

• Tiny (512×512): Fast processing, 64 vision tokens
• Small (640×640): Balanced speed/quality, 100 vision tokens
• Base (1024×1024): High quality, 256 vision tokens
• Large (1280×1280): Maximum quality, 400 vision tokens
• Gundam (dynamic): Adaptive multi-tile processing for large documents

Dataset Structure

The dataset contains all original columns plus:

• markdown: The extracted text in markdown format with preserved structure
• inference_info: JSON list tracking all OCR models applied to this dataset

Usage

from datasets import load_dataset
import json

# Load the dataset
dataset = load_dataset("{{output_dataset_id}}", split="train")

# Access the markdown text
for example in dataset:
    print(example["markdown"])
    break

# View all OCR models applied to this dataset
inference_info = json.loads(dataset[0]["inference_info"])
for info in inference_info:
    print(f"Column: {{info['column_name']}} - Model: {{info['model_id']}}")

Reproduction

This dataset was generated using the uv-scripts/ocr DeepSeek OCR script:

uv run https://huggingface.co/datasets/uv-scripts/ocr/raw/main/deepseek-ocr.py \
    Alysonhower/test \
    <output-dataset> \
    --resolution-mode gundam \
    --image-column image

Performance

• Processing Speed: ~0.0 images/second
• Processing Method: Sequential (Transformers API, no batching)

Note: This uses the official Transformers implementation. For faster batch processing, consider using the vLLM version once DeepSeek-OCR is officially supported by vLLM.

Generated with 🤖 UV Scripts