Sorry, you need to enable JavaScript to visit this website.

facebooktwittermailshare

EFFICIENT FINE-TUNING OF NEURAL NETWORKS FOR ARTIFACT REMOVAL IN DEEP LEARNING FOR INVERSE IMAGING PROBLEMS

Abstract: 

While Deep Neural Networks trained for solving inverse imaging problems (such as super-resolution, denoising, or inpainting tasks) regularly achieve new state-of-the-art restoration performance, this increase in performance is often accompanied with undesired artifacts generated in their solution. These artifacts are usually specific to the type of neural network architecture, training, or test input image used for the inverse imaging problem at hand. In this paper, we propose a fast, efficient post-processing method for reducing these artifacts. Given a test input image and its known image formation model, we fine-tune the parameters of the trained network and iteratively update them using a data consistency loss. We show that in addition to being efficient and applicable to large variety of problems, our post-processing through fine-tuning approach enhances the solution originally provided by the neural network by maintaining its restoration quality while reducing the observed artifacts, as measured qualitatively and quantitatively.

up
0 users have voted:

Paper Details

Authors:
Santiago Lopez-Tapia, Rafael Molina, Aggelos K. Katsaggelos
Submitted On:
10 September 2019 - 3:54pm
Short Link:
Type:
Poster
Event:
Document Year:
2019
Cite

Document Files

ICIP_2019_Efficient_Finetuning_v2.pdf

(29)

Subscribe

[1] Santiago Lopez-Tapia, Rafael Molina, Aggelos K. Katsaggelos, "EFFICIENT FINE-TUNING OF NEURAL NETWORKS FOR ARTIFACT REMOVAL IN DEEP LEARNING FOR INVERSE IMAGING PROBLEMS", IEEE SigPort, 2019. [Online]. Available: http://sigport.org/4573. Accessed: Dec. 13, 2019.
@article{4573-19,
url = {http://sigport.org/4573},
author = {Santiago Lopez-Tapia; Rafael Molina; Aggelos K. Katsaggelos },
publisher = {IEEE SigPort},
title = {EFFICIENT FINE-TUNING OF NEURAL NETWORKS FOR ARTIFACT REMOVAL IN DEEP LEARNING FOR INVERSE IMAGING PROBLEMS},
year = {2019} }
TY - EJOUR
T1 - EFFICIENT FINE-TUNING OF NEURAL NETWORKS FOR ARTIFACT REMOVAL IN DEEP LEARNING FOR INVERSE IMAGING PROBLEMS
AU - Santiago Lopez-Tapia; Rafael Molina; Aggelos K. Katsaggelos
PY - 2019
PB - IEEE SigPort
UR - http://sigport.org/4573
ER -
Santiago Lopez-Tapia, Rafael Molina, Aggelos K. Katsaggelos. (2019). EFFICIENT FINE-TUNING OF NEURAL NETWORKS FOR ARTIFACT REMOVAL IN DEEP LEARNING FOR INVERSE IMAGING PROBLEMS. IEEE SigPort. http://sigport.org/4573
Santiago Lopez-Tapia, Rafael Molina, Aggelos K. Katsaggelos, 2019. EFFICIENT FINE-TUNING OF NEURAL NETWORKS FOR ARTIFACT REMOVAL IN DEEP LEARNING FOR INVERSE IMAGING PROBLEMS. Available at: http://sigport.org/4573.
Santiago Lopez-Tapia, Rafael Molina, Aggelos K. Katsaggelos. (2019). "EFFICIENT FINE-TUNING OF NEURAL NETWORKS FOR ARTIFACT REMOVAL IN DEEP LEARNING FOR INVERSE IMAGING PROBLEMS." Web.
1. Santiago Lopez-Tapia, Rafael Molina, Aggelos K. Katsaggelos. EFFICIENT FINE-TUNING OF NEURAL NETWORKS FOR ARTIFACT REMOVAL IN DEEP LEARNING FOR INVERSE IMAGING PROBLEMS [Internet]. IEEE SigPort; 2019. Available from : http://sigport.org/4573