update yaml
Browse files
README.md
CHANGED
|
@@ -1,3 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
# ClimateGAN: Raising Awareness about Climate Change by Generating Images of Floods
|
| 2 |
|
| 3 |
This repository contains the code used to train the model presented in our **[paper](https://arxiv.org/abs/2110.02871)**.
|
|
@@ -20,7 +32,7 @@ If you use this code, data or pre-trained weights, please cite our ICLR 2022 pap
|
|
| 20 |
|
| 21 |
## Using pre-trained weights
|
| 22 |
|
| 23 |
-
In the paper, we present ClimateGAN as a solution to produce images of floods. It can actually do **more**:
|
| 24 |
|
| 25 |
* reusing the segmentation map, we are able to isolate the sky, turn it red and in a few more steps create an image resembling the consequences of a wildfire on a neighboring area, similarly to the [California wildfires](https://www.google.com/search?q=california+wildfires+red+sky&source=lnms&tbm=isch&sa=X&ved=2ahUKEwisws-hx7zxAhXxyYUKHQyKBUwQ_AUoAXoECAEQBA&biw=1680&bih=917&dpr=2).
|
| 26 |
* reusing the depth map, we can simulate the consequences of a smog event on an image, scaling the intensity of the filter by the distance of an object to the camera, as per [HazeRD](http://www2.ece.rochester.edu/~gsharma/papers/Zhang_ICIP2017_HazeRD.pdf)
|
|
@@ -54,11 +66,11 @@ If you want to use Comet, make sure you have the [appropriate configuration in p
|
|
| 54 |
$ cd config
|
| 55 |
$ gdown https://drive.google.com/u/0/uc?id=18OCUIy7JQ2Ow_-cC5xn_hhDn-Bp45N1K
|
| 56 |
$ unzip release-github-v1.zip
|
| 57 |
-
$ cd ..
|
| 58 |
```
|
| 59 |
|
| 60 |
2. Run from the repo's root:
|
| 61 |
-
|
| 62 |
1. With `comet`:
|
| 63 |
|
| 64 |
```bash
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language:
|
| 3 |
+
- en
|
| 4 |
+
tags:
|
| 5 |
+
- Climate Change
|
| 6 |
+
- GAN
|
| 7 |
+
- Domain Adaptation
|
| 8 |
+
license: gpl-3.0
|
| 9 |
+
# datasets:
|
| 10 |
+
# -
|
| 11 |
+
---
|
| 12 |
+
|
| 13 |
# ClimateGAN: Raising Awareness about Climate Change by Generating Images of Floods
|
| 14 |
|
| 15 |
This repository contains the code used to train the model presented in our **[paper](https://arxiv.org/abs/2110.02871)**.
|
|
|
|
| 32 |
|
| 33 |
## Using pre-trained weights
|
| 34 |
|
| 35 |
+
In the paper, we present ClimateGAN as a solution to produce images of floods. It can actually do **more**:
|
| 36 |
|
| 37 |
* reusing the segmentation map, we are able to isolate the sky, turn it red and in a few more steps create an image resembling the consequences of a wildfire on a neighboring area, similarly to the [California wildfires](https://www.google.com/search?q=california+wildfires+red+sky&source=lnms&tbm=isch&sa=X&ved=2ahUKEwisws-hx7zxAhXxyYUKHQyKBUwQ_AUoAXoECAEQBA&biw=1680&bih=917&dpr=2).
|
| 38 |
* reusing the depth map, we can simulate the consequences of a smog event on an image, scaling the intensity of the filter by the distance of an object to the camera, as per [HazeRD](http://www2.ece.rochester.edu/~gsharma/papers/Zhang_ICIP2017_HazeRD.pdf)
|
|
|
|
| 66 |
$ cd config
|
| 67 |
$ gdown https://drive.google.com/u/0/uc?id=18OCUIy7JQ2Ow_-cC5xn_hhDn-Bp45N1K
|
| 68 |
$ unzip release-github-v1.zip
|
| 69 |
+
$ cd ..
|
| 70 |
```
|
| 71 |
|
| 72 |
2. Run from the repo's root:
|
| 73 |
+
|
| 74 |
1. With `comet`:
|
| 75 |
|
| 76 |
```bash
|