accessibility object detection Computer Vision Project

In this data annotation project, I am focusing on training accessibility features such as ramps, potholes, accessibility symbols, persons, and stairs. The objective of the project is to create a high-quality, annotated dataset that can be used for training an object detection model to recognize these important features in various environments.

The source of the images used for this project includes diverse datasets such as Kaggle, Google Images, and the COCO dataset. These sources provide a wide variety of scenarios and settings, ensuring that the annotated data encompasses different lighting conditions, angles, and backgrounds.

The annotation process involves meticulously marking the accessibility features in each image, ensuring precise bounding boxes and class labels for accurate training data. This careful attention to detail will contribute to developing a robust and reliable model that can accurately identify and classify accessibility features in real-world scenarios.

By completing this data annotation project, I aim to contribute to the development of models that can enhance accessibility and safety in public spaces and urban environments.

Data Sources:

1. https://www.kaggle.com/datasets/nderalparslan/dwsonder Alparslan, Ö., Özcan, A. & Çetin, Ö. (2022). Model ve Veri Odaklı Yaklaşımlar ile Nesne Tespit Başarısının Arttırılmasına Yönelik ESA ve Veri Seti Optimizasyonları . Mühendislik Bilimleri ve Araştırmaları Dergisi , 4 (2) , 121-128 . DOI: 10.46387/bjesr.1106501

2. Google images

3. https://cocodataset.org/