barbarametzler/clustering_satellite_images_Accra

repository for the publication titled "Phenotyping urban built and natural environments with high-resolution satellite images and unsupervised deep learning" # Clustering satellite images of Accra Code used for the paper "Phenotyping urban built and natural environments with high-resolution satellite images and unsupervised deep learning" published in STOTEN, 2023. Access here: doi.org The clustering algorithm was adapted from the DeepCluster algorithm by Caron (2018), which was published by Facebook research and is also openly available, was run on 3 RTX6000 GPUs, 72GB memory and a runtime of approximately 24 h. github.com ## Requirements - Python 3 - the SciPy and scikit-learn packages - a PyTorch install version 0.1.8 (pytorch.org) - CUDA 8.0 - a Faiss install (Faiss) - The ImageNet dataset (which can be automatically downloaded by recent version of torchvision) ## Usage as defined in satcluster/DeepCluster/example_script.sh ``` main_sfeatures.py [-h] [--arch ARCH] [--sobel] [--clustering {Kmeans,PIC}] [--nmb_cluster NMB_CLUSTER] [--lr LR] [--wd WD] [--reassign REASSIGN] [--workers WORKERS] [--epochs EPOCHS] [--start_epoch START_EPOCH] [--batch BATCH] [--momentum MOMENTUM] [--resume PATH] [--checkpoints CHECKPOINTS] [--seed SEED] [--exp EXP] [--features_epoch] [--features_epoch] [--verbose] DIR positional arguments: DIR path to dataset optional arguments: -h, --help show this help message and exit --arch ARCH, -a ARCH11CNN architecture (default: vgg16) --sobel Sobel filtering --clustering {Kmeans,PIC} clustering algorithm (default: Kmeans) --nmb_cluster NMB_CLUSTER, --k NMB_CLUSTER number of cluster for k-means (default: 8) --lr LR learning rate (default: 0.0001) --wd WD weight decay pow (default: -5) --reassign REASSIGN how many epochs of training between two consecutive reassignments of clusters (default: 1) --workers WORKERS number of data loading workers (default: 4) --epochs EPOCHS number of total epochs to run (default: 20) --start_epoch START …