Open AccessA Survey On Auto-Image Colorization Using Deep Learning Techniques With User Proposition
Author(s) -
C. Santhanakrishnan,
Neeraj Durgapal,
Deepak Yadav
Publication year - 2019
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1362/1/012094
Subject(s) - computer science , artificial intelligence , convolutional neural network , deep learning , grayscale , scale (ratio) , computer vision , image (mathematics) , set (abstract data type) , convolution (computer science) , data set , artificial neural network , dependency (uml) , pattern recognition (psychology) , geography , cartography , programming language
An approach based on deep learning for automatic colorization of image with optional user-guided hints. The system maps a gray-scale image, along with, user hints” (selected colors) to an output colorization with a Convolution Neural Network (CNN). Previous approaches have relied heavily on user input which results in non-real-time desaturated outputs. The network takes user edits by fusing low-level information of source with high-level information, learned from large-scale data. Some networks are trained on a large data set to eliminate this dependency. The image colorization systems find their applications in astronomical photography, CCTV footage, electron microscopy, etc. The various approaches combine color data from large data sets and user inputs provide a model for accurate and efficient colorization of grey-scale images.