Retrieve the characters in an image using EasyOCR Optical Character Recognition (OCR).
This block returns the text within an image.
You may want to use this block in combination with a detections-based block (i.e. ObjectDetectionBlock). An object detection model could isolate specific regions from an image (i.e. a shipping container ID in a logistics use case) for further processing. You can then use a DynamicCropBlock to crop the region of interest before running OCR.
Using a detections model then cropping detections allows you to isolate your analysis on particular regions of an image.
Note that EasyOCR has limitations running within containers on Apple Silicon.
Type identifier
Use the following identifier in step "type" field: roboflow_core/easy_ocr@v1 to add the block as a step in your workflow.
Properties
| Name | Type | Description | Refs |
|---|---|---|---|
name | str | Unique name of step in workflows. | ❌ |
language | str | Language model to use for OCR. | ❌ |
quantize | bool | Quantized models are smaller and faster, but may be less accurate and won't work correctly on all hardware.. | ❌ |
The Refs column marks possibility to parametrise the property with dynamic values available in workflow runtime. See Bindings for more info.
Input and Output Bindings
The available connections depend on its binding kinds. Check what binding kinds EasyOCR in version v1 has.
Input and output bindings
input
images(image): The image to infer on..
output
result(string): String value.predictions(object_detection_prediction): Prediction with detected bounding boxes in form of sv.Detections(...) object.parent_id(parent_id): Identifier of parent for step output.root_parent_id(parent_id): Identifier of parent for step output.prediction_type(prediction_type): String value with type of prediction.
Example JSON definition
{
"name": "<your_step_name_here>",
"type": "roboflow_core/easy_ocr@v1",
"images": "$inputs.image",
"language": "<block_does_not_provide_example>",
"quantize": "<block_does_not_provide_example>"
}