infer_deepseek_ocr

DeepSeek-OCR by DeepSeek AI, use groundbreaking approach to compressing long contexts via optical 2D mapping. This innovative system demonstrates that vision-based compression can achieve remarkable efficiency in handling text-heavy documents, potentially revolutionizing how large language models (LLMs) process extensive textual information.

The DeepSeek-OCR system consists of two primary components: DeepEncoder and DeepSeek3B-MoE-A570M as the decoder. Together, they achieve an impressive 97% OCR precision when compressing text at a ratio of less than 10× (meaning 10 text tokens compressed into 1 vision token).

🚀 Use with Ikomia API

1. Install Ikomia API

We strongly recommend using a virtual environment. If you're not sure where to start, we offer a tutorial here.

pip install ikomia

2. Create your workflow

from ikomia.dataprocess.workflow import Workflow
from ikomia.utils.displayIO import display

# Init your workflow
wf = Workflow()

# Add algorithm
algo = wf.add_task(name="infer_deepseek_ocr", auto_connect=True)

# Run on your image  
wf.run_on(url="https://github.com/NanoNets/Nanonets-OCR2/blob/main/assets/bank_statement.jpg?raw=true")

# Display input
display(algo.get_input(0).get_image())
# Save output .json
deepseek_output = algo.get_output(1)
deepseek_output.save('deepseek_output.json')

☀️ Use with Ikomia Studio

Ikomia Studio offers a friendly UI with the same features as the API.

• If you haven't started using Ikomia Studio yet, download and install it from this page.
• For additional guidance on getting started with Ikomia Studio, check out this blog post.

📝 Set algorithm parameters

• model_name (string, default: deepseek-ai/DeepSeek-OCR): Hugging Face model repo to load.
• cuda (bool, default: auto): Use GPU if available. If set to True but no CUDA device is present, it will fall back to CPU.
• prompt (string, default: "<|grounding|>Convert the document to markdown."): Text instruction appended after the image token to control the output style and task.
• mode (enum, default: Gundam): Preset controlling resolution and cropping. One of: Tiny, Small, Base, Large, Gundam.
  • Gundam (Recommended): Balanced performance with crop mode # base_size = 1024, image_size = 640, crop_mode = True
  • Base: Standard quality without cropping # base_size = 1024, image_size = 1024, crop_mode = False
  • Large: Highest quality for complex documents # base_size = 1280, image_size = 1280, crop_mode = False
  • Small: Faster processing, good for simple text # base_size = 640, image_size = 640, crop_mode = False
  • Tiny: Fastest, suitable for clear printed text # base_size = 512, image_size = 512, crop_mode = False
• test_compress (bool, default: True): Enable internal compression/fast path to reduce compute and VRAM. Turn off for maximum fidelity.

from ikomia.dataprocess.workflow import Workflow
from ikomia.utils.displayIO import display

# Init your workflow
wf = Workflow()

# Add algorithm
algo = wf.add_task(name="infer_deepseek_ocr", auto_connect=True)

algo.set_parameters({
    'prompt': "<|grounding|>Convert the document to markdown.",
    'mode': "Gundam",
    'test_compress': "True",
    })

# Run on your image  
wf.run_on(url="https://github.com/NanoNets/Nanonets-OCR2/blob/main/assets/bank_statement.jpg?raw=true")

# Show input
display(algo.get_input(0).get_image())
# Save output .json
deepseek_output = algo.get_output(1)
deepseek_output.save('deepseek_output.json')

🔍 Explore algorithm outputs

Every algorithm produces specific outputs, yet they can be explored them the same way using the Ikomia API. For a more in-depth understanding of managing algorithm outputs, please refer to the documentation.

from ikomia.dataprocess.workflow import Workflow

# Init your workflow
wf = Workflow()

# Add algorithm
algo = wf.add_task(name="infer_deepseek_ocr", auto_connect=True)

# Run on your image  
wf.run_on(url="https://github.com/NanoNets/Nanonets-OCR2/blob/main/assets/bank_statement.jpg?raw=true")

# Iterate over outputs
for output in algo.get_outputs():
    # Print information
    print(output)
    # Export it to JSON
    output.to_json()

⏩ Advanced usage

💡 Tips for Best Results

• For receipts: Use "ocr" mode with "gundam" or "base" preset
• For documents with tables: Use "markdown" mode with "large" preset
• If text is not detected: Try different presets in this order: gundam → base → large
• For handwritten text: Use "large" preset for better accuracy
• Ensure images are clear and well-lit for optimal results

📝 Prompts examples

• document: <|grounding|>Convert the document to markdown.
• other image: <|grounding|>OCR this image.
• without layouts: Free OCR.
• figures in document: Parse the figure.
• general: Describe this image in detail.
• rec: Locate <|ref|>xxxx<|/ref|> in the image.
• '先天下之忧而忧'