Cycle-Spinning GAN for Raindrop Removal from Images

Date

2019-09-21

Author

Uzun, Ülkü
Temizel, Alptekin

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

230
views

0
downloads

Weather events such as rain, snow, and fog degrade the quality of images taken under these conditions. Enhancement of such images is critical for intelligent transport and outdoor surveillance systems. Generative Adversarial Networks (GAN) based methods have been shown to be promising for enhancing these images in recent years. In this study, we adapt the cycle-spinning technique to GAN for removal of raindrops. The experimental evaluation of the proposed method shows that the performance is improved in terms of reference-based metrics (SSIM and PSNR). In addition, the approach also results in higher object detection performance in terms of mean average precision (mAP) metric when applied before the detection process.

Subject Keywords

Quality assessment, Measurement, Object detection, Generative adversarial networks, Spinning, Training, Gallium nitride, Rain

URI

https://hdl.handle.net/11511/31515

DOI

https://doi.org/10.1109/avss.2019.8909824

Collections

Graduate School of Informatics, Conference / Seminar

Suggestions

OpenMETU
Core

Towards 5G and beyond radio link diagnosis: Radio link failure prediction by using historical weather, link parameters Aktaş, Semih; Alemdar, Hande; Ergüt, Salih (2022-04-01) Weather-related phenomena such as clouds, rain, snow affect the performance of radio links. To reduce the adverse effects of radio link failures’ on the user experience, mobile operators require intelligent monitoring systems to predict link failures and take actions before they happen. In this study, we show how machine learning can be used for prediction using a real-world telecom operator dataset. We propose a novel architecture to process time-series data and non-times-series data together in the same n...
Assessment of flash flood events using remote sensing and atmospheric model-derived precipitation in a hydrological model Yücel, İsmail (2011-07-07) Remotely-sensed precipitation estimates and regional atmospheric model precipitation forecasts provide rainfall data at high spatial and temporal resolutions with a large-scale coverage, and can therefore be potentially used for hydrological applications for making flash flood forecasts and warnings. This study investigates the performance of the rainfall products obtained from the Hydro Estimator (HE) algorithm of NOAA/NESDIS and the Weather Research and Forecasting (WRF) model, and their use in a hydrolog...
Assessment of a flash flood event using different precipitation datasets Yücel, İsmail (2015-12-01) Remotely sensed precipitation estimates and regional atmospheric model precipitation forecasts provide rainfall data at high spatial and temporal resolutions and can therefore be potentially used for hydrological applications for flash flood forecasting and warning. This study investigates the performance of the rainfall products obtained from weather radar, the Hydro-Estimator (HE) algorithm of NOAA/NESDIS and the Weather Research and Forecasting (WRF) model, and their use in the Hydrologic Engineering Cen...
Monitoring the snow covered Areas from NOAA AVHRR images in the Eastern Part of Turkey Akyürek, Sevda Zuhal (2002-04-01) Monitoring snow-covered areas and estimating the snow water equivalent play an important role in predicting discharges during spring months, especially in regions where snow is an important resource.^This study has been conducted in the Upper Euphrates River basin, of 10 200 knr area, and elevation range of 1125— 3500 m. In estimating snow-covered areas, besides semi-supervised multispectral classification of NOAA-AVHRR data, a theta algorithm, developed by the US National Weather Service, has been used. Th...
Evaluating the use of radar precipitation data in simulating a flood event in Terme Samsun Turkey Akyürek, Sevda Zuhal (2016-10-10) Floods caused by convective storms in mountainous regions are sensitive to the temporal and spatial variability of precipitation. In developing countries, researches and authorities working on this issue have to deal with lack of data. Even a dense raingauge network may not be able to state variability of the precipitation patterns especially in mountainous regions. Space-time estimates of rainfall from weather radar can be a remedy to represent pattern of the rainfall with some inaccuracy. In the field of ...

Citation Formats

Ü. Uzun and A. Temizel, “Cycle-Spinning GAN for Raindrop Removal from Images,” 2019, Accessed: 00, 2020. [Online]. Available: https://hdl.handle.net/11511/31515.