0.25
0.5
0.75
1.25
1.5
1.75
2
Learning Recursive Filters for Low-Level Vision via a Hybrid Neural Network
Published on Oct 24, 20163301 Views
In this paper, we consider numerous low-level vision problems (e.g., edge-preserving filtering and denoising) as recursive image filtering via a hybrid neural network. The network contains several spa
Related categories
Chapter list
Learning Recursive Filters for Low-Level Vision via a Hybrid Neural Network00:00
Introduction00:09
Low-Level Vision Problems: Filtering00:51
Low-Level Vision Problems: Enhancement00:59
Low-Level Vision Problems: Image Denoising01:02
Low-Level Vision Problems: Image Inpainting01:05
Low-Level Vision Problems: Color Interpolation01:08
Contributions01:12
Convolutional Filter/101:34
Convolutional Filter/202:06
Recursive Filter/102:09
Recursive Filter/202:30
Recursive Filter/302:42
Framework of Hybrid Network/103:15
Framework of Hybrid Network/204:47
Perspective from Signal Processing/104:58
Perspective from Signal Processing/205:11
Perspective from Signal Processing/305:20
Perspective from Signal Processing/405:22
Perspective from Signal Processing/505:36
Perspective from Signal Processing/606:01
Perspective from Signal Processing/706:16
Spatially Variant Linear RNN/106:35
Spatially Variant Linear RNN/207:04
Hybrid Network: Joint Training/107:12
Hybrid Network: Joint Training/207:29
Hybrid Network: Joint Training/307:37
Hybrid Network: Joint Training/407:43
Hybrid Network: Linear RNNs/107:48
Hybrid Network: Linear RNNs/207:59
Hybrid Network: Linear RNNs/308:05
Hybrid Network: Linear RNNs/408:09
Hybrid Network: CNN 08:19
Model Stability08:39
Weight Maps with Single LRNN/109:07
Weight Maps with Single LRNN/209:30
Low-Level Vision Tasks09:39
Edge-Preserving Smoothing10:08
Edge-Preserving Smoothing: Rolling Guidance Filter10:34
Edge-Preserving Enhancement: Shock Filter10:50
Image Denoising11:04
Image Pixel Propagation: 50% Random Pixels11:17
Image Pixel Propagation: Character Inpainting11:30
Color Pixel Propagation: 3% Color Retained/111:40
Color Pixel Propagation: 3% Color Retained/211:53
Re-colorization/112:06
Re-colorization/212:24
Run Time and Model Size12:33
Concluding Remarks12:49
Demo: Cartooning13:09