Update README.md

README.md

@@ -12,33 +12,25 @@ base_model:
 - facebook/nougat-small
 ---
 
-# Arabic Small Nougat: End-to-End Structured OCR for Arabic Books
-
-**Arabic Small Nougat** is an end-to-end Optical Character Recognition (OCR) model designed specifically for the Arabic language. It converts images of Arabic book pages into structured text, with output formatted in Markdown. The model is based on the [facebook/nougat-small](https://huggingface.co/facebook/nougat-small) architecture and fine-tuned on the [Khatt dataset](https://huggingface.co/datasets/Fakhraddin/khatt), with additional custom data to enhance performance for Arabic text recognition.
+# Arabic Small Nougat 
 
+**En**d-**t**o-**En**d **Structur**ed **OC**R **fo**r **Arab**ic **boo**ks.
 <center>
   <img src="https://huggingface.co/MohamedRashad/arabic-small-nougat/resolve/main/thumbnail_image.jpg">
 </center>
 
 ## Description
 
-The **Arabic Small Nougat** OCR model is designed to perform high-quality, end-to-end text extraction from images of Arabic book pages. Fine-tuned from the Nougat model architecture, it is tailored to handle the nuances of Arabic scripts, including their rich structure and complex typography.
-
-- **Key Features:**
-  - Support for both **Arabic** and **English** text
-  - **Markdown-formatted output** ideal for digital book projects, academic work, and content conversion
-  - Capable of handling full-page book scans, producing well-structured text
-  - Built on a robust OCR pipeline, ideal for Arabic literature and document digitization
+The arabic-small-nougat OCR is an end-to-end structured Optical Character Recognition (OCR) system designed specifically for the Arabic language.
 
-## Getting Started
+The model is based on the [facebook/nougat-small](https://huggingface.co/facebook/nougat-small) architecture and has been fine-tuned using the [Khatt dataset](https://huggingface.co/datasets/Fakhraddin/khatt) along with a custom dataset created for this purpose.
 
-**Demo:** [Live Demo](https://huggingface.co/spaces/MohamedRashad/Arabic-Small-Nougat)
+## How to Get Started with the Model
 
-**Quickstart Code:**
+**Demo:** https://huggingface.co/spaces/MohamedRashad/Arabic-Small-Nougat
 
-Here’s a code snippet to get started with the model:
-
-```python
+Or, use the code below to get started with the model locally.
+python
 from PIL import Image
 import torch
 from transformers import NougatProcessor, VisionEncoderDecoderModel
@@ -49,104 +41,78 @@ model = VisionEncoderDecoderModel.from_pretrained("MohamedRashad/arabic-small-no
 device = "cuda" if torch.cuda.is_available() else "cpu"
 model.to(device)
 
-# Define maximum context length for generated text
 context_length = 2048
 
 def predict(img_path):
-    """
-    Predicts OCR text from an image of a book page.
-    """
-    # Open and preprocess the image
+    # prepare PDF image for the model
     image = Image.open(img_path)
     pixel_values = processor(image, return_tensors="pt").pixel_values
-    
-    # Generate transcription
+
+    # generate transcription
     outputs = model.generate(
         pixel_values.to(device),
         min_length=1,
-        max_new_tokens=context_length
+        max_new_tokens=context_length,
+        bad_words_ids=[[processor.tokenizer.unk_token_id]],
     )
-    
-    # Decode the generated text and post-process
+
     page_sequence = processor.batch_decode(outputs, skip_special_tokens=True)[0]
     page_sequence = processor.post_process_generation(page_sequence, fix_markdown=False)
-    
     return page_sequence
 
-# Example usage
 print(predict("path/to/page_image.jpg"))
-```
-
-### Installation Requirements
-
-Make sure to install the following dependencies:
-```bash
-pip install -U transformers Pillow torch
-```
-
 ## Bias, Risks, and Limitations
 
-- **Text Hallucination**: The model may occasionally generate repetitive or incorrect text due to the complexities of OCR tasks, especially with noisy or difficult-to-read images.
-- **Image Quality Sensitivity**: Low-resolution or blurred images may significantly affect the accuracy of the OCR.
-- **Context Length Limitation**: The model has a maximum context length of 2048 tokens, which may result in incomplete transcriptions for long book pages or documents with extensive content.
+1. **Text Hallucination:** The model may occasionally generate repeated or incorrect text due to the inherent complexities of OCR tasks.
+1. **Erroneous Image Paths:** There are instances where the model outputs image paths that are not relevant to the input, indicating occasional confusion.
+1. **Context Length Constraint:** The model has a maximum context length of 2048 tokens, which may result in incomplete transcriptions for longer book pages.
 
 ## Intended Use
 
-This model is ideal for converting scanned or photographed images of Arabic book pages into digital text, making it useful for:
-- **Digitizing Arabic literature**
-- **Extracting structured text from scanned books**
-- **Academic research** and archival purposes, particularly in the humanities and social sciences
+The arabic-small-nougat OCR is designed for tasks that involve converting images of Arabic book pages into structured text, especially when Markdown format is desired. It is suitable for applications in the field of digitizing Arabic literature and facilitating text extraction from printed materials.
 
 ## Ethical Considerations
 
-- **Accuracy Verification**: Users should always verify the output for accuracy, especially for critical applications or when processing historical texts.
-- **Copyright and Fair Use**: This tool can be used for digitizing public domain works, but care should be taken when dealing with copyrighted material. Ensure proper permissions and comply with intellectual property laws when using this model in such contexts.
+It is crucial to be aware of the model's limitations, particularly in instances where accurate OCR results are critical. Users are advised to verify and review the output, especially in scenarios where precision is paramount.
 
 ## Model Details
 
-- **Developed by**: Mohamed Rashad
-- **Model Type**: VisionEncoderDecoderModel
-- **Languages**: Arabic, English
-- **License**: GPL 3.0
-- **Base Model**: [facebook/nougat-small](https://huggingface.co/facebook/nougat-small)
-- **Fine-Tuned Dataset**: [Khatt dataset](https://huggingface.co/datasets/Fakhraddin/khatt)
+- **Developed by:** Mohamed Rashad
+- **Model type:** VisionEncoderDecoderModel
+- **Language(s) (NLP):** Arabic & English
+- **License:** GPL 3.0
+- **Finetuned from model:** [nougat-small](https://huggingface.co/facebook/nougat-small)
 
-### Acknowledgments
+## Acknowledgment
 
-If you use or build upon the **Arabic Small Nougat** OCR model, please acknowledge the contributions of:
-- The original authors of the [facebook/nougat-small](https://huggingface.co/facebook/nougat-small) model
-- The creators of the [Khatt dataset](https://huggingface.co/datasets/Fakhraddin/khatt)
-- The open-source community and the model’s developer, Mohamed Rashad
+If you use or build upon the Arabic Small Nougat OCR, please acknowledge the model developer and the open-source community for their contributions. Additionally, be sure to include a copy of the GPL 3.0 license with any redistributed or modified versions of the model.
 
-### Citation
+By selecting the GPL 3.0 license, you promote the principles of open source and ensure that the benefits of the model are shared with the broader community.
 
-If you find this model useful, please cite the following references:
+### Citation
 
-```bibtex
+If you find this model useful, please consider citing the original [facebook/nougat-small]((https://huggingface.co/facebook/nougat-small)) model and the datasets used for fine-tuning, including the [Khatt dataset](https://huggingface.co/datasets/Fakhraddin/khatt) and any details regarding the custom dataset.
+bibtex
 @misc{blecher2023nougat,
-  title={Nougat: Neural Optical Understanding for Academic Documents}, 
-  author={Lukas Blecher and Guillem Cucurull and Thomas Scialom and Robert Stojnic},
-  year={2023},
-  eprint={2308.13418},
-  archivePrefix={arXiv},
-  primaryClass={cs.LG}
+      title={Nougat: Neural Optical Understanding for Academic Documents}, 
+      author={Lukas Blecher and Guillem Cucurull and Thomas Scialom and Robert Stojnic},
+      year={2023},
+      eprint={2308.13418},
+      archivePrefix={arXiv},
+      primaryClass={cs.LG}
 }
-
 @misc{fakhraddin2023khatt,
-  title={Khatt Arabic Handwriting Dataset},
-  author={Fakhraddin},
-  year={2023},
-  howpublished={\url{https://huggingface.co/datasets/Fakhraddin/khatt}}
+      title={Khatt Arabic Handwriting Dataset},
+      author={Fakhraddin},
+      year={2023},
+      howpublished={\url{https://huggingface.co/datasets/Fakhraddin/khatt}}
 }
-
 @misc{rashad2023arabicsmallnougat,
-  title={Arabic Small Nougat Model},
-  author={Mohamed Rashad},
-  year={2023},
-  howpublished={\url{https://huggingface.co/MohamedRashad/arabic-small-nougat}}
+      title={Arabic Small Nougat Model},
+      author={Mohamed Rashad},
+      year={2023},
+      howpublished={\url{https://huggingface.co/MohamedRashad/arabic-small-nougat}}
 }
-```
-
-## Disclaimer
+### Disclaimer
 
-The **Arabic Small Nougat OCR** model is provided "as is." The developers make no guarantees regarding its suitability for specific tasks. Users are encouraged to evaluate the model’s performance and carefully review the output for their specific use cases and requirements.
+The arabic-small-nougat OCR is a tool provided "as is," and the developers make no guarantees regarding its suitability for specific tasks. Users are encouraged to thoroughly evaluate the model's output for their particular use cases and requirements.