Karen Petrosyan, Narek Okroyan

Deep Learning Based Old Photo Restoration and Colorization Web Service


Karen Petrosyan is a Machine Learning and Computer Vision Engineer at Pixeria Lab, holding a Bachelor’s degree in Computer Science from the American University of Armenia. His experience includes designing and training deep learning models for image classification and semantic segmentation tasks. He is also involved in research as a Machine Learning Researcher at FAST Foundation. With a background in machine learning, computer vision, and image processing, he is passionate about utilizing these technologies to solve real-world problems and contribute to the advancement of the field.



Narek Okroyan is a web developer specializing in the Node.js + React stack. He holds a Bachelor’s degree in Computer Science from the American University of Armenia. Throughout his career, he has worked on various projects that demonstrate his expertise in web development. Notably, he has developed an online video editor with integrated machine learning features and a web scraper service, enabling automated data extraction and analysis from various websites. While web development is his primary focus, his passion for image processing and machine learning has always been a driving force in his work. He is enthusiastic about exploring the intersection of these fields and leveraging their potential to solve complex problems.


Description of the Talk:

The talk discusses comprehensive approach for restoring and colorizing old photographs using neural networks and image processing techniques. The objective was to develop a user-friendly web service that allows effortless enhancement and modification of images. By leveraging convolutional neural networks and deep learning techniques, there was trained a model capable of restoring damaged images. The successful deployment of the models on the web service demonstrates their capabilities and offers a convenient tool for users without technical expertise.