Image enhancement on low-light and dark images for object detection using Artificial Intelligence for field practitioners
Vivek Kumar Varma Nadimpalli, Gopichand Agnihotram · Institution of Engineering and Technology eBooks · 2022
In recent times, there is a lot of demand for Artificial Intelligence solutions based on computer vision in various fields. Many solutions like object detection, fault detection, environment description, and scene prediction are helping to solve many real-life problems. But these solutions are dependent on vision-based computations. Generally, all these computations are designed in such a way that environment in each frame is visible and computation performed with data captured from the frame is in visible condition. But in the case of dark condition, the photons count that a camera capture decreases drastically, and the environment may not be visible. In this scenario, the system will fail to compute all the tasks that are dependent on visibility of environment. With the increase in Artificial Intelligence solutions using vision data, it is important to process low-light/dark images and draw intelligence from them. The information and subsequent intelligence available during low-light scenarios can be extracted wisely by our proposed deep learning architecture. The algorithm will process the raw data taken from the camera sensor and provides you the enhanced JPEG images. These enhanced images will be used to train the object detection using TensorFlow lite to detect the objects in the frames. The entire solution will be ported into the mobile devices for capturing the raw data to enhance images and detect the objects on the enhanced images. The proposed chapter will also explain how this solution will be used in the field assistance where user can be able to see the objects in the scene clearly with the enhanced images and detect the objects for machine repair and maintenance of various tasks.