8K Arabic OCR Dataset for OCR Detection and Recognition Training

arabic OCR dataset

arabic text recognition dataset

scene text dataset

OCR training data

text detection dataset

This dataset contains 8,604 images collected from diverse real-world Arabic text scenes, covering various environments, shooting angles, and natural conditions. Each text instance is annotated with precise quadrilateral or polygon bounding boxes along with transcription labels, enabling accurate text localization and recognition. This data can be used for Arabic OCR applications and multilingual computer vision model training.

This is a paid datasets for commercial use, research purpose and more. Licensed ready made datasets help jump-start AI projects.

Specifications

Data size

8,604 images, 65,231 Arabic quadrilateral bounding boxes, 909 Arabic polygon bounding boxes

Collecting environment

including shop plaque, stop board, poster, road sign, comic, prompt/reminder, warning, packing instruction, menu, building sign, magazine book covers, etc.

Data diversity

including a variety of natural scenes, multiple shooting angles

Device

cellphone, camera

Photographic angle

looking up angle, looking down angle, eye-level angle

Data format

the image data format is .jpg, the annotation file format is .json

Annotation content

line-level quadrilateral bounding box annotation and transcription for the texts, polygon bounding box annotation and transcription for the texts

Accuracy

the error bound of each vertex of quadrilateral or polygon bounding box is within 5 pixels, which is a qualified annotation, the accuracy of bounding boxes is not less than 95%; the texts transcription accuracy is not less than 95%

Recommended Dataset

English OCR Dataset in Natural Scenes – 71,535 Images

This dataset contains 71,535 English natural-scene images collected from real environments in the UK and the United States. The data diversity includes multiple scenes, multiple photographic angles and multiple light conditions. For annotation, line-level & word-leve & character-level rectangular bounding box or quadrilateral bounding box annotation were adopted, the text transcription was also adopted. This dataset is suitable for English OCR, scene text detection, and text recognition research in real-world environments.

English OCR dataset scene text dataset street sign text dataset outdoor OCR dataset image text dataset text detection dataset text recognition dataset

500,000 Images – Multilingual OCR Dataset in 21 Languages

This dataset covers 21 languages, with 20,000 to 25,000 images per language. The data includes natural scenes, document photography scenes, and electronic scenes. The data diversity includes various data types, multiple shooting angles, and multiple languages. In terms of annotation, quadrilateral or polygonal at the row (column) level and content transcription at the row (column) level are adopted. This dataset can be use for multilingual optical character recognition (OCR) and text detection tasks.

multilingual OCR dataset scene text recognition data document OCR dataset electronic screen OCR data OCR dataset 21 languages AI OCR training data text recognition dataset

426,687 Images - Multilingual OCR Dataset – Document & Scene Text

426,687 high-resolution images featuring multilingual Optical Character Recognition (OCR) data across both natural scenes and various document types. This dataset spans 20 languages, including Traditional Chinese, Simplified Chinese, Japanese, Korean, Thai, Vietnamese, Indonesian, Malay, Polish, and more. The data covers a wide range of real-world conditions—natural scenes, printed documents, handwritten notes, signs, and posters—captured from multiple countries and environments, with varied backgrounds, lighting conditions, and camera angles. All images are annotated for OCR tasks, making this dataset highly suitable for training deep learning models for text detection, recognition, and layout analysis in multi-language scenarios. The dataset complies with global data protection standards (GDPR, CCPA, PIPL), and is validated by leading AI enterprises for commercial and research applications.

multilingual OCR dataset scene text dataset document OCR images Chinese OCR Japanese OCR dataset Thai text recognition OCR dataset with annotation multi-language OCR document image dataset OCR training data

29,954 Images - Southeast Asian OCR Dataset (Khmer, Lao, Burmese)

29,954 Images - OCR Collection Data of Southeast Asian Languages, including Khmer (Cambodia), Lao and Burmese. The diversity of collection includes multiple languages, multiple collection types, multiple shooting angles. This set of data can be used for Southeast Asian language OCR tasks.

Southeast Asian OCR dataset Khmer OCR data Lao OCR dataset Burmese OCR dataset natural scene OCR minority language text recognition multilingual OCR dataset

Turkish OCR Dataset – 5,000 Natural Scene Images

This dataset contains 5,000 images of Turkish natural scenes with text.This data include a variety of natural scenarios and multiple shooting angles. For annotation, quadrilateral or polygon bounding box annotation and transcription for the texts were annotated in the data. This data can be used for Turkish OCR, scene text recognition, and text detection in natural images.

Turkish OCR dataset Turkish text recognition dataset Turkish scene text dataset Turkish natural scene OCR dataset Turkish document OCR dataset

Large Korean & Hindi Scene Text OCR Dataset – 104,000+ Natural Images

This dataset contains 104,320 images of Korean and Hindi text in natural scenes. The collecting scenes include packaging, posters, tickets, reminders, menus, building signs, and similar real-world environments. The dataset features diverse scenes, multiple shooting angles and multiple lighting conditions. For annotation, line-level polygon bounding box (or tetragon or rectangular bounding boxes) annotation, transcription and text attributes (language type) for the texts; vertical-level polygon bounding box (or tetragon or rectangular bounding boxes) annotation, transcription and text attributes (language type) for the text. The dataset can be used for Korean and Hindi OCR tasks in natural scenes.

Korean OCR dataset Hindi OCR dataset Scene text OCR dataset Multilingual OCR images OCR training data for AI

Vertical Text OCR Dataset with 57,645 Images and Polygon Annotations

This dataset contains 57,645 images collected from diverse real-world text scenes, including street scenes, shop signs, billboards, posters, decorative text, art lettering, and magazine covers. The dataset includes Chinese text and a small amount of English text. Each image is annotated with text localization information and transcription labels, including polygon and quadrilateral bounding boxes for vertical and non-vertical text regions. This dataset is suitable for tasks such as scene text OCR tasks and multi-oriented text recognition.

ocr dataset ocr training data scene text dataset text recognition dataset text detection dataset

Multilingual OCR Dataset – 12 Languages Natural Scene Text

This dataset contains 105,941 images of natural scene text collected across multiple real-world environments, covering 12 languages, including 6 Asian languages and 6 European languages. The data covers multiple natural scenes, multiple photographic angles. For annotation, each image is annotated with line-level quadrilateral bounding boxes and accurate text transcriptions. The data can be used for multilingual OCR, scene text recognition, and cross-language OCR model training.

multilingual OCR dataset OCR dataset scene text OCR dataset natural scene OCR dataset multi-language OCR dataset text recognition dataset OCR annotation dataset line-level OCR dataset

8K Arabic OCR Dataset for OCR Detection and Recognition Training

arabic OCR dataset arabic text recognition dataset scene text dataset OCR training data text detection dataset

Current Project Maturity

arabic OCR dataset

arabic text recognition dataset

scene text dataset

OCR training data

text detection dataset