Benchmarking OCR APIs on Actual-World Paperwork

With the speedy developments in Massive Language Fashions (LLMs) and Imaginative and prescient-Language Fashions (VLMs), many consider OCR has turn out to be out of date. If LLMs can “see” and “learn” paperwork, why not use them straight for textual content extraction?

The reply lies in reliability. Are you able to all the time be a 100% certain of the veracity of textual content output that LLMs interpret from a doc/picture? We put this to check with a easy experiment. We requested colleagues to make use of any LLM of their option to extract a listing of passenger names (10) from a pattern PDF flight ticket.

Whereas LLMs can interpret and summarize paperwork, they lack the precision and structured output required for important enterprise functions the place 100% information accuracy is essential. Moreover, LLMs require important computational sources, making them expensive and impractical for large-scale doc processing, particularly in enterprise and edge deployments.

OCR, however, is optimized for effectivity, working on low-power units whereas delivering constant outcomes. When accuracy is non-negotiable whether or not in monetary data, authorized contracts, or regulatory compliance, OCR stays probably the most reliable answer.

In contrast to LLMs, OCR APIs present confidence scores and bounding containers, permitting builders to detect uncertainties in extracted textual content. This degree of management is essential for companies that can’t afford incorrect or hallucinated information. That’s why OCR APIs proceed to be broadly utilized in doc automation workflows, AI-driven information extraction, and enterprise functions.

To evaluate the state of OCR in 2025, we benchmarked 9 of the most well-liked OCR APIs, protecting business options, open-source OCR engines, and doc processing frameworks. Our purpose is to offer an goal, data-driven comparability that helps builders and enterprises select the perfect device for his or her wants.

Methodology

Dataset Choice:

To make sure a complete analysis of OCR APIs or OCR fashions in real-world eventualities, we chosen datasets that embody a various vary of doc sorts and challenges generally encountered in sensible functions. Our dataset decisions embody:

• Frequent Enterprise Paperwork: Varieties, invoices, and monetary statements containing structured textual content.
• Receipts: Printed transaction slips with various fonts, noise, and light textual content.
• Low-Decision Photographs: Paperwork captured beneath suboptimal situations, mimicking real-world scanning and pictures limitations.
• Handwritten Textual content: Samples with completely different handwriting kinds to check handwriting recognition capabilities.
• Blurred or Distorted Textual content: Photographs with movement blur or compression artifacts to evaluate OCR robustness.
• Rotated or Skewed Textual content: Paperwork scanned or photographed at an angle, requiring superior textual content alignment dealing with.
• Tabular Knowledge: Paperwork containing structured tabular data, difficult for OCR fashions to protect format integrity.
• Dense Textual content: Textual content-heavy paperwork, similar to account opening types, to judge efficiency in high-content areas.

To make sure our benchmark covers all these real-world challenges, we choose the next datasets:

1. STROIE (hyperlink to dataset)

2. FUNSD (hyperlink to dataset)

These datasets present a complete testbed for evaluating OCR efficiency throughout sensible and actual life eventualities.

Fashions Choice

To guage OCR efficiency throughout completely different eventualities, we embody a mixture of business APIs, open-source OCR fashions, and doc processing frameworks. This ensures a balanced comparability between proprietary options and freely obtainable options. The fashions utilized in our benchmark are:

• In style Industrial OCR APIs:
  • Google Cloud Imaginative and prescient AI
  • Azure AI Doc Intelligence
  • Amazon Textract
• In style Open-Supply OCR APIs:
  • Surya
  • PaddleOCR
  • RapidOCR
  • Extractous
• In style Open-Supply Doc Processing Frameworks:

To reveal how every OCR API processes a picture, we offer code snippets for working OCR utilizing each business APIs and open-source frameworks. These examples present find out how to load a picture, apply OCR, and extract the textual content, providing a sensible information for implementation and comparability. Beneath are the code snippets for every mannequin:

1. Google Cloud Imaginative and prescient AI: First step is to arrange a brand new Google Cloud Mission. Within the Google Cloud Console, navigate to APIs & Providers → Library, seek for Imaginative and prescient API, and click on Allow. Go to APIs & Providers → Credentials, click on Create Credentials → Service Account, identify it (e.g., vision-ocr-service), and click on Create & Proceed. Assign the Proprietor (or Editor) function and click on Achieved. Now, in Service Accounts, choose the account, go to Keys → Add Key → Create New Key, select JSON, and obtain the .json file.

Required Packages:

pip set up google-cloud-vision

from google.cloud import imaginative and prescient
from google.oauth2 import service_account

credentials = service_account.Credentials.from_service_account_file("/content material/ocr-nanonets-cea4ddeb1dd2.json") #path to the json file downloaded

shopper = imaginative and prescient.ImageAnnotatorClient(credentials=credentials)

def detect_text(image_path):
   """Detects textual content in a picture utilizing Google Cloud Imaginative and prescient AI."""
   with open(image_path, 'rb') as image_file:
       content material = image_file.learn()

   picture = imaginative and prescient.Picture(content material=content material)
   response = shopper.text_detection(picture=picture)
  
   texts = response.text_annotations

   if texts:
       return texts[0].description
   else:
       return "No textual content detected."

   if response.error.message:
       increase Exception(f"Error: {response.error.message}")

# Exchange along with your picture path
image_path = "/content material/drive/MyDrive/OCR_datasets/STROIE/test_data/img/X00016469670.jpg"
print(detect_text(image_path))

2. Azure AI Doc Intelligence: Create an Azure Account (Azure Portal) to get $200 free credit for 30 days. Within the Azure Portal, go to Create a Useful resource, seek for Azure AI Doc Intelligence (Type Recognizer), and click on Create. Select a Subscription, Useful resource Group, Area (nearest to you), set Pricing Tier to Free (if obtainable) or Customary, then click on Assessment + Create → Create. As soon as created, go to the Azure AI Doc Intelligence useful resource, navigate to Keys and Endpoint, and replica the API Key and Endpoint.

Required Packages:

pip set up azure-ai-documentintelligence

from azure.ai.documentintelligence import DocumentIntelligenceClient
from azure.core.credentials import AzureKeyCredential
import io
# Exchange along with your Azure endpoint and API key
AZURE_ENDPOINT = "https://your-region.api.cognitive.microsoft.com/"
AZURE_KEY = "your-api-key"

shopper = DocumentIntelligenceClient(AZURE_ENDPOINT, AzureKeyCredential(AZURE_KEY))
def extract_text(image_path):
   """Extracts textual content from a picture utilizing Azure AI Doc Intelligence."""
   with open(image_path, "rb") as image_file:
       image_data = image_file.learn()
       
   poller = shopper.begin_analyze_document("prebuilt-read", doc=image_data)
   
   consequence = poller.consequence()
   extracted_text = []
   for web page in consequence.pages:
       for line in web page.strains:
           extracted_text.append(line.content material)

   print("Detected textual content:")
   print("n".be a part of(extracted_text))
   
image_path = image_path
extract_text(image_path)

3. Amazon Textract: Create an AWS Account (AWS Signal-Up) to entry Amazon Textract’s free-tier (1,000 pages/month for 3 months). Within the AWS Administration Console, go to IAM (Id & Entry Administration) → Customers → Create Consumer, identify it (e.g., textract-user), and choose Programmatic Entry. Below Permissions, connect AmazonTextractFullAccess and AmazonS3ReadOnlyAccess (if utilizing S3). Click on Create Consumer and replica the Entry Key ID and Secret Entry Key.

Required Packages:

pip set up boto3

Set Atmosphere Variables:

export AWS_ACCESS_KEY_ID="your-access-key"
export AWS_SECRET_ACCESS_KEY="your-secret-key"
export AWS_REGION="your-region"

import boto3
textract = boto3.shopper("textract", region_name="us-east-1")

def extract_text(image_path):
   """Extracts textual content from a picture utilizing Amazon Textract."""
   with open(image_path, "rb") as image_file:
       image_bytes = image_file.learn()

   response = textract.detect_document_text(Doc={"Bytes": image_bytes})
   extracted_text = []
   for merchandise in response["Blocks"]:
       if merchandise["BlockType"] == "LINE":
           extracted_text.append(merchandise["Text"])
   print("Detected textual content:")
   print("n".be a part of(extracted_text))
   
image_path = image_path
extract_text(image_path)

4. Surya :  Use pip set up surya-ocr to obtain the required packages. Then create a python file with the next code and run it in terminal.

from PIL import Picture
from surya.recognition import RecognitionPredictor
from surya.detection import DetectionPredictor

picture = Picture.open(image_path)
langs = ["en"]

recognition_predictor = RecognitionPredictor()
detection_predictor = DetectionPredictor()

predictions = recognition_predictor([image], [langs], detection_predictor)

5. PaddleOCR : Use “pip set up paddleocr paddlepaddle” to put in the required packages. Then create a python file with the next code and run it in terminal.

from paddleocr import PaddleOCR

ocr = PaddleOCR(use_angle_cls=True, lang="en")

consequence = ocr.ocr(image_path, cls=True)

6. RapidOCR : Use “pip set up rapidocr_onnxruntime” to put in the required packages. Then create a python file with the next code and run it in terminal.

from rapidocr_onnxruntime import RapidOCR

engine = RapidOCR()

img_path = image_path
consequence, elapse = engine(img_path)

7. Extractous: Use “sudo apt set up tesseract-ocr tesseract-ocr-deu” to put in the required packages. Then create a python file with the next code and run it in terminal.

from extractous import Extractor, TesseractOcrConfig
extractor = Extractor().set_ocr_config(TesseractOcrConfig().set_language("en"))

consequence, metadata = extractor.extract_file_to_string(image_path)

print(consequence)

8. Marker: Use “pip set up marker-pdf” to put in the required packages. Then in terminal use the next code.

!marker_single image_path --output_dir saving_directory --output_format json

9. Unstructured-IO: Use “pip set up “unstructured[image]”” to put in the required packages. Then create a python file with the next code and run it in terminal.

from unstructured.partition.auto import partition

components = partition(filename=image_path)

print("nn".be a part of([str(el) for el in elements]))

Analysis Metrics

To evaluate the effectiveness of every OCR mannequin, we consider each accuracy and efficiency utilizing the next metrics:

• Character Error Price (CER): Measures the ratio of incorrect characters (insertions, deletions, and substitutions) to the full characters within the floor fact textual content. Decrease CER signifies higher accuracy.
• Phrase Error Price (WER): Just like CER however operates on the phrase degree, calculating errors relative to the full variety of phrases. It helps assess how effectively fashions acknowledge full phrases.
• ROUGE Rating: A textual content similarity metric that compares OCR output with the bottom fact primarily based on overlapping n-grams, capturing each precision and recall.

For efficiency analysis, we measure:

• Inference Time (Latency per Picture): The time taken by every mannequin to course of a single picture, indicating pace and effectivity in real-world functions.

Price Analysis:

• For business OCR APIs, price is set by their pricing fashions, usually primarily based on the variety of processed pages or pictures. 
• For open-source OCR APIs, whereas there are not any direct utilization prices, we assess computational overhead by measuring reminiscence utilization throughout inference.

Benchmarking Outcomes

Because the datasets used—STROIE (completely different receipt pictures) and FUNSD (enterprise paperwork with tabular layouts)—include numerous format kinds, the extracted textual content varies throughout fashions primarily based on their skill to protect construction. This variation impacts the Phrase Error Price (WER) and Character Error Price (CER), as these metrics rely upon the place of phrases and characters within the output.

A excessive error charge signifies {that a} mannequin struggles to take care of the chronological order of textual content, particularly in advanced layouts and tabular codecs.

1. Phrase Error Price

WER of every mannequin on the FUNSD and STROIE datasets is introduced under. These outcomes spotlight how effectively every mannequin preserves phrase order throughout completely different doc layouts.

2. Character Error Price

CER of every mannequin on the FUNSD and STROIE datasets is introduced under. These outcomes point out how precisely every mannequin captures character-level textual content whereas dealing with completely different doc layouts.

Why are the WER and CER metrics for Surya and Marker so excessive on the STROIE dataset?

STROIE’s intricate layouts make OCR troublesome. Surya tries to fill gaps by inserting further phrases, resulting in excessive WER and CER, even after post-processing. Marker, which makes use of Surya for OCR and outputs markdown textual content, inherits these points. The markdown formatting additional misaligns textual content, worsening the error charges.

Variation in Mannequin Efficiency Throughout Datasets

OCR fashions carry out otherwise primarily based on dataset construction. Google Cloud Imaginative and prescient AI and Azure AI Doc Intelligence deal with numerous layouts higher, whereas open-source fashions like RapidOCR and Surya battle with structured codecs, resulting in extra errors.

Because the fashions battle to protect layouts, resulting in excessive WER and CER, we use one other metric—ROUGE Rating—to evaluate textual content similarity between the mannequin’s output and the bottom fact. In contrast to WER and CER, ROUGE focuses on content material similarity quite than phrase place. Which means that no matter format preservation, a excessive ROUGE rating signifies that the extracted textual content intently matches the reference, whereas a low rating suggests important content material discrepancies.

3. ROUGE Rating

ROUGE Rating of every mannequin on the FUNSD and STROIE datasets is introduced under. These outcomes mirror the content material similarity between the extracted textual content and the bottom fact, no matter format preservation.

The ROUGE scores reveal that Google Cloud Imaginative and prescient AI constantly outperforms different fashions throughout each FUNSD (75.0%) and STROIE (87.8%), indicating superior textual content extraction. Surya and Marker, which depend on the identical backend, present comparable efficiency, although Marker barely lags on STROIE (70.3%). Extractous and Unstructured-IO rating the bottom in each datasets, suggesting weaker textual content coherence. PaddleOCR and Azure AI Doc Intelligence obtain balanced outcomes, making them aggressive options. The general development highlights the energy of business APIs, whereas open-source fashions exhibit combined efficiency.

4. Latency per picture

Latency per picture for every mannequin is introduced under. This measures the time taken by every mannequin to carry out OCR on one picture, offering insights into their effectivity and processing pace.

The latency evaluation exhibits that Google Cloud Imaginative and prescient AI, Amazon Textract, and Extractous keep a very good steadiness between pace and accuracy. Surya and Paddle OCR exhibit notably greater inference instances, suggesting potential inefficiencies. Open-source fashions like Fast OCR and Marker fluctuate in efficiency, with some providing aggressive speeds whereas others lag behind. Azure AI Doc Intelligence additionally exhibits average latency, making it a viable selection relying on the use case.

5. Price or reminiscence utilization per picture

For business APIs, we current the utilization price (price per 1000 pictures processed). For open-source fashions, the metric signifies reminiscence consumption as a proxy for price, offering insights into their useful resource effectivity.

Amongst open-source fashions, Marker and Unstructured-IO have considerably greater reminiscence consumption, which can impression deployment in resource-constrained environments. Surya and Extractous strike a steadiness between efficiency and reminiscence effectivity. PaddleOCR and RapidOCR are probably the most light-weight choices, making them ultimate for low-memory eventualities.

Conclusion

Primarily based on the analysis throughout latency, inference time, and ROUGE rating, no single mannequin dominates in all features. Nonetheless, some fashions stand out in particular areas:

• Greatest Latency & Inference Time: Extractous and Amazon Textract reveal the quickest response instances, making them ultimate for real-time functions.
• Greatest ROUGE Rating (Accuracy): Google Cloud Imaginative and prescient AI and Azure AI Doc Intelligence obtain the best accuracy in textual content recognition, making them sturdy candidates for functions requiring exact OCR.
• Greatest Reminiscence Effectivity: RapidOCR and PaddleOCR devour the least reminiscence, making them extremely appropriate for low-resource environments.

Greatest Mannequin General

Contemplating a steadiness between accuracy, pace, and effectivity, Google Cloud Imaginative and prescient AI emerges as the perfect total performer. It offers sturdy accuracy with aggressive inference time. Nonetheless, for open sourced fashions, PaddleOCR and RapidOCR supply the perfect trade-off between accuracy, pace and reminiscence effectivity.

Leaderboard of Greatest OCR APIs primarily based on completely different efficiency metrics:

LLM vs. Devoted OCR: A Case Examine

To know how OCR fashions evaluate to Massive Language Fashions (LLMs) in textual content extraction, we examined a difficult picture utilizing each LLaMa 3.2 11B Imaginative and prescient  and RapidOCR, a small however devoted OCR mannequin.

Outcomes:

1. LLaMa 3.2 11B Imaginative and prescient
   • Struggled with faint textual content, failing to reconstruct sure phrases.
   • Misinterpreted some characters and added hallucinated phrases.
   • Took considerably longer to course of the picture.
   • Used a variety of compute sources.
2. RapidOCR
   • Precisely extracted a lot of the textual content regardless of the troublesome situations.
   • Ran effectively on very low compute sources.

Is OCR Nonetheless Related In the present day?

With the rise of multimodal LLMs able to decoding pictures and textual content, some consider OCR might turn out to be out of date. Nonetheless, the truth is extra nuanced.

Should you or your finish prospects have to be 100% certain of information you are extracting from paperwork or pictures, OCR nonetheless is your finest guess for now! Confidence scores and bounding containers from OCR APIs can be utilized to deduce when the output shouldn’t be dependable.

With LLMs you may by no means be 100% certain of the veracity of the textual content output due to hallucinations and the insecurity scores.

Who Nonetheless Wants OCR?

• Enterprises Dealing with Excessive-Quantity Paperwork: Banks, authorized corporations, and insurance coverage corporations depend on OCR for automated doc processing at scale.
• Governments and Compliance: Passport scanning, tax data, and regulatory filings nonetheless require OCR for structured extraction.
• AI-Powered Knowledge Pipelines: Many companies combine OCR with NLP pipelines to transform paperwork into structured information earlier than making use of AI fashions.
• Multilingual and Low-Useful resource Language Functions: OCR stays important for digitizing uncommon scripts the place LLMs lack coaching information.

Why Ought to Enterprises Nonetheless Care About OCR When Everybody Desires LLMs?

1. Accuracy and Reliability: LLMs generate hallucinations, whereas OCR ensures exact textual content extraction, making it important for authorized, monetary, and authorities functions.
2. Pace and Price Effectivity: OCR is light-weight and works on edge units, whereas LLMs require excessive compute sources and cloud inference prices.
3. The longer term shouldn’t be OCR vs. LLMs—it’s OCR and LLMs: OCR can extract clear textual content, and LLMs can then course of and interpret it for insights. AI-powered OCR fashions will proceed to enhance, integrating LLM reasoning for higher post-processing.

Last Ideas

Whereas LLMs have expanded the probabilities of textual content extraction from pictures, OCR stays indispensable for structured, high-accuracy textual content retrieval and can all the time be essential for dependable doc processing. Moderately than changing OCR, LLMs will complement it, bringing higher understanding, context, and automation to extracted information.