RainGAN: unsupervised raindrop removal via decomposition and composition
Adherent raindrops on windshield or camera lens may distort and occlude vision, causing issues for downstream machine vision perception. Most of the existing raindrop removal methods focus on learning the mapping from a raindrop image to its clean content by training with the paired raindrop-clean i...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis-Master by Research |
| Language: | English |
| Published: |
Nanyang Technological University
2022
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/160029 |
| _version_ | 1811695663839182848 |
|---|---|
| author | Xu, Yan |
| author2 | Loke Yuan Ren |
| author_facet | Loke Yuan Ren Xu, Yan |
| author_sort | Xu, Yan |
| collection | NTU |
| description | Adherent raindrops on windshield or camera lens may distort and occlude vision, causing issues for downstream machine vision perception. Most of the existing raindrop removal methods focus on learning the mapping from a raindrop image to its clean content by training with the paired raindrop-clean images. However, the paired real-world images are difficult to collect in practice. This thesis presents a novel framework for raindrop removal that eliminates the need for paired training samples. Based on the assumption that a raindrop image is the composition of a clean image and a raindrop style, the proposed framework decomposes a raindrop image into a clean content image and a raindrop-style latent code and composes a clean content image and a raindrop style code to a raindrop image for data augmentation. The proposed framework introduces a domain-invariant residual block to facilitate the identity mapping for the clean portion of the raindrop image. Extensive experiments on real-world raindrop datasets show that our network can achieve superior performance in raindrop removal to other unpaired image-to-image translation methods, even with comparable performance with state-of-the-art methods that require paired images. |
| first_indexed | 2024-10-01T07:27:03Z |
| format | Thesis-Master by Research |
| id | ntu-10356/160029 |
| institution | Nanyang Technological University |
| language | English |
| last_indexed | 2024-10-01T07:27:03Z |
| publishDate | 2022 |
| publisher | Nanyang Technological University |
| record_format | dspace |
| spelling | ntu-10356/1600292022-08-01T05:07:19Z RainGAN: unsupervised raindrop removal via decomposition and composition Xu, Yan Loke Yuan Ren School of Computer Science and Engineering NCS Pte Ltd yrloke@ntu.edu.sg Engineering::Computer science and engineering Adherent raindrops on windshield or camera lens may distort and occlude vision, causing issues for downstream machine vision perception. Most of the existing raindrop removal methods focus on learning the mapping from a raindrop image to its clean content by training with the paired raindrop-clean images. However, the paired real-world images are difficult to collect in practice. This thesis presents a novel framework for raindrop removal that eliminates the need for paired training samples. Based on the assumption that a raindrop image is the composition of a clean image and a raindrop style, the proposed framework decomposes a raindrop image into a clean content image and a raindrop-style latent code and composes a clean content image and a raindrop style code to a raindrop image for data augmentation. The proposed framework introduces a domain-invariant residual block to facilitate the identity mapping for the clean portion of the raindrop image. Extensive experiments on real-world raindrop datasets show that our network can achieve superior performance in raindrop removal to other unpaired image-to-image translation methods, even with comparable performance with state-of-the-art methods that require paired images. Master of Engineering 2022-07-12T01:46:34Z 2022-07-12T01:46:34Z 2022 Thesis-Master by Research Xu, Y. (2022). RainGAN: unsupervised raindrop removal via decomposition and composition. Master's thesis, Nanyang Technological University, Singapore. https://hdl.handle.net/10356/160029 https://hdl.handle.net/10356/160029 10.32657/10356/160029 en This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC 4.0). application/pdf Nanyang Technological University |
| spellingShingle | Engineering::Computer science and engineering Xu, Yan RainGAN: unsupervised raindrop removal via decomposition and composition |
| title | RainGAN: unsupervised raindrop removal via decomposition and composition |
| title_full | RainGAN: unsupervised raindrop removal via decomposition and composition |
| title_fullStr | RainGAN: unsupervised raindrop removal via decomposition and composition |
| title_full_unstemmed | RainGAN: unsupervised raindrop removal via decomposition and composition |
| title_short | RainGAN: unsupervised raindrop removal via decomposition and composition |
| title_sort | raingan unsupervised raindrop removal via decomposition and composition |
| topic | Engineering::Computer science and engineering |
| url | https://hdl.handle.net/10356/160029 |
| work_keys_str_mv | AT xuyan rainganunsupervisedraindropremovalviadecompositionandcomposition |