The official codes for paper "Deep hash learning for remote sensing image retrieval"

numpy
opencv-python
torch
torchvision

We conduct the experiments on three data sets, including UC Merced, AID, and NWPU-RESISC45. To train and test our model, you should download the data set and modify each image's path in the dataset/AID/.txt or dataset/NWPU/.txt or dataset/UC_Merced/.txt (depending which data set you select to conduct the experiment)

All the configurations are in trainerAndHash.py, and you can modify them by your needs.

python trainerAndHash.py --phase=0

python trainerAndHash.py --phase=1

the path of codes can be modified the line about "parser.add_argument('--codes_dir', default=root + '/codes', type=str)" in trainerAndHash.py

python trainerAndHash.py --phase=2