Auto-Colorization of Historical Images Using Deep Convolutional Neural Networks

Engineering & Materials Science

• Deep neural networks 100%
• Convolutional neural networks 93%
• Deep learning 61%
• Signal to noise ratio 56%
• Backpropagation 32%
• Computer vision 29%
• Fusion reactions 27%
• Image processing 26%
• Machine learning 25%
• Color 24%
• Neural networks 20%

Mathematics

• Neural Networks 96%
• Signal-to-noise Ratio 32%
• Mean Squared Error 25%
• Cultural Heritage 24%
• User Studies 24%
• Learning 23%
• Quality Assessment 20%
• Back Propagation 19%
• Local Features 19%
• Color Image 18%
• Feedforward 17%
• Culture 15%
• Computer Vision 15%
• Enhancement 15%
• Image Processing 14%
• Machine Learning 14%
• Fusion 13%
• Training 13%
• Truth 13%
• Model 12%
• Predict 11%
• Human 11%
• Objective function 10%