Landmark detection and recognition using Google's Landmark Detection V2 dataset. This dataset contains 5 million images, split into directories, and a train.csv. Since it is impossible for my computer to be able to train a model for the galore images present, the code takes very few images from three directories.
To download the images, follow https://s3.amazonaws.com/google-landmark/train/images_000.tar this convention. You may even download 104, 056, dpending on your personal preference.
Examplar images would be:
