EasyOCR#

EasyOCR is a ready-to-use OCR with 80+ supported languages. It helps you to quickly convert and transcribe text from images. This guide provides an overiew of using EasyOCR with BentoML.

Compatibility#

BentoML has been validated to work with EasyOCR version 1.6.2 and higher.

Save/Load a EasyOCR Reader with BentoML#

First, create a reader instance with the language codes for your usecase.

import easyocr

 reader = easyocr.Reader(['en'])

Save this reader instance using save_model() to save this to the BentoML model store

import bentoml

bento_model = bentoml.easyocr.save_model('en-reader', reader)

To verify that the saved model is working, load it back with load_model():

loaded_model = bentoml.easyocr.load_model('en-reader')

rs = loaded_model.readtext('image.jpg')

Note

GPU behaviour

GPU can be passed through easyocr.Reader constructor as gpu=True. This means in order to use GPU, the reader instance must be created with a machine with GPU before saving it to BentoML.

Building a Service#

See also

Building a Service: more information on creating a prediction service with BentoML.

Create a service.py file separate from your training code that will be used to define the BentoML service:

import bentoml
import PIL.Image
import numpy as np

# create a runner from the saved Booster
runner = bentoml.easyocr.get("en-reader").to_runner()

# create a BentoML service
svc = bentoml.Service("ocr", runners=[runner])

# define a new endpoint on the BentoML service
@svc.api(input=bentoml.io.Image(), output=bentoml.io.JSON())
async def transcript_text(input: PIL.Image.Image) -> list:
    # use 'runner.predict.run(input)' instead of 'booster.predict'
    return await runner.readtext.async_run(np.asarray(input))

Take note of the name of the service (svc in this example) and the name of the file.

You should also have a bentofile.yaml alongside the service file that specifies that information, as well as the fact that it depends on XGBoost. This can be done using either python (if using pip), or conda:

pip

service: "service:svc"
python:
  packages:
     - easyocr
     - bentoml

conda

service: "service:svc"
conda:
  channels:
  - conda-forge
  dependencies:
  - easyocr

EasyOCR#

Compatibility#

Save/Load a EasyOCR Reader with BentoML#

Building a Service#

Using Runners#