Please use this identifier to cite or link to this item:
http://20.198.91.3:8080/jspui/handle/123456789/8733| Title: | Hyper realistic rain video simulation and study on deep learning based deraining |
| Authors: | Chanda, Aishik |
| Advisors: | Maulik, Ujjwal |
| Keywords: | Hyper Realistic Rain;Deep Learning |
| Issue Date: | 2022 |
| Publisher: | Jadavpur University, Kolkata, West Bengal |
| Abstract: | Raindrops adhered to windshields of cars can severely hamper the visibility of the background and degrade any image or video captured by the driving assistance systems to an extent that it no longer is able to detect background objects accurately. The problem is intractable since the regions occluded by the raindrops are not given and since any information about the occluded regions are completely lost. It is difficult to develop any supervised model as no dataset with proper ground truth exists for rainy videos. In this paper we attempt to develop a method to restore the video frame by frame by first creating a dataset of rainy images with their ground truth and then developing a model to restore rainy images. |
| URI: | http://20.198.91.3:8080/jspui/handle/123456789/8733 |
| Appears in Collections: | Dissertations |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| M.E. (Computer Science and Engineering) Aishik Chanda.pdf | 4.3 MB | Adobe PDF | View/Open |
Items in IR@JU are protected by copyright, with all rights reserved, unless otherwise indicated.