surya-ocr 0.1.0

Document OCR models for multilingual text detection and recognition

Surya

Surya is a multilingual document OCR toolkit. It can do:

• Accurate line-level text detection
• Line-level text recognition (coming soon)
• Table and chart detection (coming soon)

It works on a range of documents and languages (see usage and benchmarks for more details).

Surya is named after the Hindu sun god, who has universal vision.

Community

Discord is where we discuss future development.

Examples

Name	Text Detection
New York Times	Image
Japanese	Image
Chinese	Image
Hindi	Image
Presentation	Image
Scientific Paper	Image
Scanned Document	Image

Installation

You'll need python 3.9+ and PyTorch. You may need to install the CPU version of torch first if you're not using a Mac or a GPU machine. See here for more details.

Install with:

pip install surya-ocr

Model weights will automatically download the first time you run surya.

Usage

• Inspect the settings in surya/settings.py. You can override any settings with environment variables.
• Your torch device will be automatically detected, but you can override this. For example, TORCH_DEVICE=cuda. Note that the mps device has a bug (on the Apple side) that may prevent it from working properly.

Text line detection

You can detect text lines in an image or pdf with the following command. This will write out a json file with the detected bboxes, and optionally save images of the pages with the bboxes.

Setting the DETECTOR_BATCH_SIZE env var properly will make a big difference when using a GPU. Each batch item will use 280MB of VRAM, so very high batch sizes are possible. Depending on your CPU core count, DETECTOR_BATCH_SIZE might make a difference there too.

surya_detect PDF_PATH --images

• --images will save images of the pages and detected text lines (optional)
• --max specifies the maximum number of pages to process if you don't want to process everything
• --results_dir specifies the directory to save results to instead of the default

This has worked with every language I've tried. It will work best with documents, and may not work well with photos or other images. It will also not work well with handwriting.

You can adjust DETECTOR_NMS_THRESHOLD and DETECTOR_TEXT_THRESHOLD if you don't get good results. Try lowering them to detect more text, and vice versa.

Importing in Python

You can also do text detection from code with:

from PIL import Image
from surya.detection import batch_inference
from surya.model.segformer import load_model, load_processor

image = Image.open(IMAGE_PATH)
model, processor = load_model(), load_processor()

# predictions is a list of dicts, one per image
predictions = batch_inference([image], model, processor)

Text recognition

Coming soon.

Table and chart detection

Coming soon.

Manual install

If you want to develop surya, you can install it manually:

• git clone https://github.com/VikParuchuri/surya.git
• cd surya
• poetry install # Installs main and dev dependencies

Limitations

• This is specialized for document OCR. It will likely not work on photos or other images. It will also not work on handwritten text.
• This will currently not detect math well. It is a limitation of the training data that is being worked on.

Benchmarks

Text line detection

Model	Time (s)	Time per page (s)	precision	recall
surya	73.468	0.286984	0.921306	0.923814
tesseract	63.2392	0.247028	0.869671	0.691873

Tesseract is CPU-based, and surya is CPU or GPU. I ran the benchmarks on a system with an A6000 GPU, and a 32 core CPU. This was the resource usage:

• tesseract - 32 CPU cores, or 8 workers using 4 cores each
• surya - 32 batch size, for 9GB VRAM usage

Methodology

Surya predicts line-level bboxes, while tesseract and others predict word-level or character-level. It's also hard to find 100% correct datasets with line-level annotations. Merging bboxes can be noisy, so I chose not to use IoU as the metric for evaluation.

I instead used coverage, which calculates:

• Precision - how well predicted bboxes cover ground truth bboxes
• Recall - how well ground truth bboxes cover predicted bboxes

First calculate coverage for each bbox, then add a small penalty for double coverage, since we want the detection to have non-overlapping bboxes. Anything with a coverage of 0.5 or higher is considered a match.

Then we calculate precision and recall for the whole dataset.

Running your own benchmarks

You can benchmark the performance of surya on your machine.

• Follow the manual install instructions above.
• poetry install --group dev # Installs dev dependencies

Text line detection

This will evaluate tesseract and surya for text line detection across a randomly sampled set of images from doclaynet.

python benchmark/detection.py --max 256

• --max controls how many images to process for the benchmark
• --debug will render images and detected bboxes
• --pdf_path will let you specify a pdf to benchmark instead of the default data
• --results_dir will let you specify a directory to save results to instead of the default one

Training

This was trained on 4x A6000s for about 5 days. It used a diverse set of 1M images as training data. It was trained from scratch using a modified segformer architecture.

Commercial usage

Text detection

The text detection model was trained from scratch, so it's okay for commercial usage. The weights are licensed cc-by-nc-sa-4.0, but I will waive that for any organization under $10M in gross revenue in the last 12 months.

If you want to remove the GPL license requirements for inference or use the weights commercially over the revenue limit, please contact me at [e-mail] for dual licensing.

Thanks

This work would not have been possible without amazing open source AI work:

• Segformer from NVIDIA
• transformers from huggingface

Thank you to everyone who makes open source AI possible.