Overview:
Colorizing black and white images is a complex task, as it involves predicting multiple color channels from a single grayscale input. This project delves into various deep learning architectures, including Convolutional Neural Networks (CNNs) and conditional Generator-Discriminator pipelines. We compare these models, assessing their performance based on well-established computer vision metrics.
Data Sources:
The project relies on grayscale images as input data, allowing us to explore the challenging world of colorization.
Tools and Technologies Used:
Key Findings:
Through extensive experimentation, we identify the optimal model for grayscale image colorization. We address the multi-modal and undetermined nature of this problem and discuss the intricacies of evaluating colorization results.
Learn More:
For a comprehensive exploration of this project, including methodologies, code, and more, please refer here.