The purpose of this project is to build a model that detects all ships in satellite images as quickly as possible and to classified small ships.
First, we need to train a CNN to detect all ships from the image. Then, we need to classified those ships to highlight small ones.
The train set is composed by 190 000 images. One image is in color and his dimension is 780x780 pixel. The test set is composed by 40 000 images in color and with the same dimension.