This is the model described in the paper QARI-OCR: High-Fidelity Arabic Text Recognition through Multimodal Large Language Model Adaptation.
This model is a fine-tuned version of unsloth/Qwen2-VL-2B-Instruct on an Arabic OCR dataset. It is optimized to perform Arabic Optical Character Recognition (OCR) for full-page text.
Code can be found at this url .
Performance is evaluated using three standard metrics:
| Model | WER ↓ | CER ↓ | BLEU ↑ |
|---|---|---|---|
| Qari-OCR-0.2.2.1-VL-2B-Instruct | 0.221 | 0.059 | 0.597 |
| AIN 8B | 0.757 | 0.309 | 0.103 |
| Qari-OCR-0.1-VL-2B-Instruct | 1.294 | 0.770 | 0.022 |
| easyOCR | 1.004 | 0.648 | 0.005 |
| pytesseract | 0.990 | 0.911 | <0.001 |
The model was trained using the following specifications:
Based on the training specifications, the model has the following limitations:
Evaluation was performed on a diverse dataset of Arabic text images, primarily featuring diacritical marks (tashkeel), measuring:
Try Qari v0.2.2.1 - Google Colab
You can load this model using the transformers and qwen_vl_utils library:
!pip install transformers qwen_vl_utils accelerate>=0.26.0 PEFT -U
!pip install -U bitsandbytes
from PIL import Image
from transformers import Qwen2VLForConditionalGeneration, AutoProcessor
import torch
import os
from qwen_vl_utils import process_vision_info
model_name = "NAMAA-Space/Qari-OCR-0.2.2.1-Arabic-2B-Instruct"
model = Qwen2VLForConditionalGeneration.from_pretrained(
model_name,
torch_dtype="auto",
device_map="auto"
)
processor = AutoProcessor.from_pretrained(model_name)
max_tokens = 2000
prompt = "Below is the image of one page of a document, as well as some raw textual content that was previously extracted for it. Just return the plain text representation of this document as if you were reading it naturally. Do not hallucinate."
image.save("image.png")
messages = [
{
"role": "user",
"content": [
{"type": "image", "image": f"file://{src}"},
{"type": "text", "text": prompt},
],
}
]
text = processor.apply_chat_template(
messages, tokenize=False, add_generation_prompt=True
)
image_inputs, video_inputs = process_vision_info(messages)
inputs = processor(
text=[text],
images=image_inputs,
videos=video_inputs,
padding=True,
return_tensors="pt",
)
inputs = inputs.to("cuda")
generated_ids = model.generate(**inputs, max_new_tokens=max_tokens)
generated_ids_trimmed = [
out_ids[len(in_ids) :] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)
]
output_text = processor.batch_decode(
generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False
)[0]
os.remove(src)
print(output_text)
This model follows the licensing terms of the original Qwen2 VL model. Please review the terms before using it commercially.
If you use this model in your research, please cite:
@article{wasfy2025qari,
title={QARI-OCR: High-Fidelity Arabic Text Recognition through Multimodal Large Language Model Adaptation},
author={Wasfy, Ahmed and Nacar, Omer and Elkhateb, Abdelakreem and Reda, Mahmoud and Elshehy, Omar and Ammar, Adel and Boulila, Wadii},
journal={arXiv preprint arXiv:2506.02295},
year={2025}
}
adapter_model.safetensors 116 MBtraining_args.bin 5.6 kB