LOW-LIGHT IMAGE ENHANCEMENT WITH RESNET ARCHITECTURE AND SELF-CALIBRATED ILLUMINATION NETWORK

Abstract

Generally, low-light image enhancement techniques are mostly not just made to achieve both visual quality and computational efficiency but also commonly invalid in unknown complex scenarios. The system is focused on the image high quality displaying of low-light images using enhancement techniques. This system is used Self Calibrated Illumination (SCI) module Network combination with Convolutional Neural Network (CNN) based on ResNet architecture Network to enhance the low-light image. In this system, Low-Light (LOL) dataset is applied. The system will be used LOL testing dataset for performance evaluation of the model. This system is implemented with the software program as Python language code and Anaconda application for running. Moreover, this system uses the three types of low-light image dataset for testing as LOL images, captured images by Camera and Black&White dataset.