Semantic segmentation
No Thumbnail Available
Date
2022
Journal Title
Journal ISSN
Volume Title
Publisher
Université Larbi Ben M'hidi Oum El Bouaghi
Abstract
Image semantic segmentation is more and more being of interest for computer vision and machine learning researchers. Many applications on the rise need accurate and efficient segmentation mechanisms: autonomous driving, indoor navigation, and even virtual or augmented reality systems to name a few. It is a building block for scene understanding by classifying all pixels of an image in a dense way; it is then possible to build abstract representations focusing on the objects and their shapes.
In this project, our work exploits a convolutional neural network (CNN) architecture "unet" for semantic segmentation; the choice of this architecture is due to its phenomenal success, it can solve the most complex problems in deep learning. It is a particularly effective tool for many types of images. Our model is trained on the Oxford-iiit pet dataset.
Description
Keywords
Computer vision, Deep learning, Semantic segmentation, Unet