| --- |
| license: other |
| task_categories: |
| - object-detection |
| - image-classification |
| language: |
| - en |
| - hi |
| tags: |
| - newspaper |
| - advertisement-detection |
| - yolo |
| - object-detection |
| - indian-newspapers |
| - computer-vision |
| - OCR |
| - document-layout |
| size_categories: |
| - 1K<n<10K |
| --- |
| |
| # Detech-AD: Indian Newspaper Advertisement Detection Dataset |
|
|
| ## Overview |
|
|
| **Detech-AD** is a dataset created for research and development of an automated advertisement detection pipeline for Indian e-newspapers. |
|
|
| The dataset contains newspaper PDFs, rendered newspaper pages, and YOLO-format annotations for advertisement detection. |
|
|
| The primary objective is to detect advertisements appearing in Indian newspapers and identify their locations on newspaper pages. |
|
|
| The dataset is intended to support computer vision, document analysis, advertisement detection, and OCR-based newspaper analysis. |
|
|
| --- |
|
|
| ## Newspapers |
|
|
| The dataset currently contains data from the following Indian newspapers: |
|
|
| - The Times of India (TOI) |
| - The Hindu |
| - Dainik Jagran |
|
|
| Additional newspapers may be added in future versions. |
|
|
| --- |
|
|
| ## Dataset Structure |
|
|
| The repository is organized as follows: |
|
|
| ```text |
| data/ |
| │ |
| ├── raw/ |
| │ ├── toi/ |
| │ ├── the_hindu/ |
| │ └── dainik_jagran/ |
| │ |
| ├── pages/ |
| │ ├── toi/ |
| │ ├── the_hindu/ |
| │ └── dainik_jagran/ |
| │ |
| └── yolo_dataset/ |
| ├── images/ |
| │ ├── train/ |
| │ ├── val/ |
| │ └── test/ |
| │ |
| └── labels/ |
| ├── train/ |
| ├── val/ |
| └── test/ |
