Post Competition Round: Completed #educational #blitz Weight: 15.0

LNDST

Detect water bodies from satellite Imagery

2605
421
33
375

Join us for the closing webinar where top participants will discuss their solutions!

🛠 Contribute: Found a typo? Or any other change in the description that you would like to see? Please consider sending us a pull request in the public repo of the challenge here.

🕵️ Introduction

Climate change is upon us.😢 How do we prepare for what is coming? With this challenge, we tackle an interesting idea that helps measure the rate of global warming and deforestation around the world. And that too through satellite images.

In this challenge, we give you satellite images to classify whether the location is of water or not?

With this challenge, we advance the Sustainable Development Goal 14 - Life below water.

Understand with code! Here is `getting started code` for you.`😄`

💾 Dataset

The Landsat dataset consists of 400x400 RGB satellite images which have been taken from the Landsat 8 satellite. In each image, there can be water and background. Your classifier should predict each pixel as `0 - background` or `1 - water` (int). In the train set, you will be given both the rgb images in jpg format and a mask where 0 will denote background, 1 will denote water. In the test set, we will only give you the RGB image.

📁 Files

Following files are available in the `resources` section:

• `train.zip` - (`1399` samples) This zip file contains all the rgb (400x400) images that can be used for training

• `train_masks.zip` - (`1399` samples) This zip file contains all the masks for the rgb images allocated for training.

• `test.zip` - (`467` samples) This zip contains rgb (400x400) images for which you will need to make the masks and submit a npy file of the same.

🚀 Submission

• Prepare a npy file which consists of flattened masks where `0` stands for `background` or `1` stands for `water` where each elements type is int8. You should iterate through image_0 to image_467 and maintain the order for evaluation to occur properly.

• Sample submission format available at sample_submission.csv in the resorces section.

Make your first submission here 🚀 !!

🖊 Evaluation Criteria

During evaluation F1 score where,

$F1 = 2 * \frac{precision*recall}{precision+recall}$