In this project, we are training a classifier to segment roads in aerial images from Google Maps, given a train dataset containing 100 labeled images of 400×400 pixels. After the training, we run a prediction on 50 test images of 608×608 pixels. The predicted images are then cropped in patches of 16×16 pixels and transformed to a .csv submission file containing the predicted output (1 for a road, 0 for background) for each patch.

An illustration of the model. Layers drawn in red are convolutional layers with stride 2, which down-sample the previous layers. Layers drawn in blue are deconvolutional layers with stride 2, which up-sample the previous layers.

A visualization of our segmentation models applied on three im- ages. From left to right, the images correspond to: input, logistic regression, naïve CNN, encoder-decoder CNN, ground-truth (if available).

An illustration of the intermediate layers in the network when predicting the segmentation a single aerial image. For each of the convolutional and deconvolutional layers, we select a slice and show it as an image here.

Link to the Kaggle competition: https://www.kaggle.com/c/epfml17-segmentation

• run.py: trains the model and computes the predictions.
• data_augmentation.py: augments the training data by rotation and cropping.
• deconvnet.py: defines the class that describes the DeconvNet model.
• helpers.py: some helper functions.
• init_archi.py: contains all the project's paths and a function to create the directories.
• mask_to_submission.py: computes the .csv to submit to Kaggle from predictions in.

To install the dependencies, run the command pip install -r requirements.txt.

To run a pretrained network, download the weights here and place them in the folder save_model. Then run the command: python3 run.py --restore_M1=True --restore_M2=True --init_archi=True.

Alexandre Carlier, Félix Schaller & Yubo Xie.