Filip Radenović

Revisiting Oxford and Paris

Data and code explaining how to load and process database and distractors images; load, crop and process query images; and evaluate three different protocols. Datasets, revisited annotations, and example features are automatically downloaded with example scripts. See the project page.

Unsupervised fine-tuning of CNN for image retrieval

Training data, fine-tuned deep neural networks, training and evaluation code to reproduce the results of our ECCV 2016 and TPAMI 2018 papers on CNN image retrieval. Code is available for both MATLAB/MatConvNet and Python/PyTorch frameworks. See the project page for more details.

Learning local feature descriptors with CNN

Implementation code in PyTorch to train and evaluate the network described in our NIPS 2017 paper on local descriptor learning. Already trained deep neural network provided as well. See the project page.

Dataset: Alps100K

Alps100K dataset contains 98,136 annotated (GPS coordinates, elevation, EXIF if available) outdoor images from mountain environments. Whenever you use the dataset, please acknowledge it by citing our BMVC 2015 paper. See the project page.