2 Hits in 2.8 sec

DANICE: Domain adaptation without forgetting in neural image compression [article]

Sudeep Katakol, Luis Herranz, Fei Yang, Marta Mrak
2021 arXiv   pre-print
In this paper, we study the adaptability of codecs to custom domains of interest. We show that NIC codecs are transferable and that they can be adapted with relatively few target domain images.  ...  However, naive adaptation interferes with the solution optimized for the original source domain, resulting in forgetting the original coding capabilities in that domain, and may even break the compatibility  ...  Adapting to new domains We introduce the problem of domain adaptation in neural image compression (DANICE), where a codec trained on a source domain X 1 is leveraged to improve compression in a target  ... 
arXiv:2104.09370v1 fatcat:nte4zosbpvedzcvgvhihlh6czm

Table of Contents

2021 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)  
A Two-Stage Deep Network for High Dynamic Range Image Reconstruction 550  ...  Adaptation without Forgetting in Neural Image Compression 1921 Sudeep Katakol (Univ. of Michigan, Ann Arbor), Luis Herranz (Computer Vision Center, UAB, Barcelona), Fei Yang (Computer Vision Center  ...  and Technology of China), and Zhibo Chen (CAS Key Laboratory of Technology in Geo-spatial Information Processing and Application System, University of Science and Technology of China) DANICE: Domain  ... 
doi:10.1109/cvprw53098.2021.00004 fatcat:yh3zw4afzneeza6jrd6377fd3y