JJEM: Special Edition 2 (December 2024)
JJEM: Special Edition 2 (December 2024)
2024-12-08
Transforming greyscale images with deep convolutional neural network
Bhagyashree, Santosh S G
The work proposed in the paper includes a high-performance colorization model for several applications, such as colorizing old photos and restoring damaged images; it is even applied in filmmaking and animation industries. Colorization does not always have the intention of re-storing an image to its exact ground truth color. Instead, though the colorization may deviate a bit from real colors, the aim is to make believable shading such that it looks pleasing aesthetically and helps in various ways to the user. We have exploited a variety of deep learning approaches, along with Convolutional Neural Networks, for the same. Using large datasets of colored images, our software enables developing and instructing deep CNNs for the extraction of relevant characteristics and the establishment of the relationships between them. The output is projected onto the problem of predicting accurate colorizations of grayscale images.
Colorization, Grayscale Images, Deep Convolutional Neural Networks, Computer Vision, Machine learning (ML)