IMPLEMENTATION OF WEB-BASED ARTIFICIAL INTELLIGENCE SYSTEM FOR IMPROVING LIGHT INTENSITY IN PHOTOS

• Main Author: Stefani, Ellena
• Format: Final Project
• Language: Indonesia
• Subjects: Teknik (Rekayasa, enjinering dan kegiatan berkaitan)
• Online Access: https://digilib.itb.ac.id/gdl/view/48046
• Institution: Institut Teknologi Bandung

Images taken with an ordinary camera including smartphone cameras in extremely low light produce dark and grainy photos. Using an ordinary photo editor like Photoshop can increase the light intensity of an image to some extent but usually loses color information and sharpness. The study was conducted to solve existing problems by creating systems that use artificial intelligence algorithms. The image will be reconstructed to be brighter and clearer by maintaining object features and color information. Therefore, in this Final Project, an artificial intelligence-based system is developed that is integrated on the web to reconstruct images from dark to bright automatically. It takes quite a large collection of light and dark image datasets for the program to work properly. This artificial intelligence system uses the TensorFlow machine learning library. The way the system works using computer vision Fully Convolutional Network that has filters at all layers and produces output in the same size. The system can be used for all variations of images such as photos of people, landscapes, and other objects. In the development of this system, the authors add a dataset in the form of photos of people in dark and bright conditions in large numbers. It is intended that the system can reconstruct portrait photos better. The output of this final project is a web created using the Flask framework. Users can upload images that want to be reconstructed on the web then the results that have been reconstructed by the artificial intelligence model will be displayed back to the web. The author gives the name CamAIra Web which is a combination of words from Camera and AI.