what is your training set and dev? - Githubissues

jandaldrop / landmark-recognition-challenge

Source code used for the google landmark recognition challenge on kaggle [19th place]

35 stars 16 forks source link

what is your training set and dev? #1

Open guyibang opened 6 years ago

guyibang commented 6 years ago

how do you get no-landmark images on train and dev?can you describe the directories in more detail?

jandaldrop commented 5 years ago

Hello and thank you for your interest!

They are constructed by this glob string non_landmark_image_files = glob.glob(non_landmark_train_path + '*.jp*g'), the path non_landmark_train_path can be set at the top of the notebook (including stars).

For the competition I have used images from the following public datasets:

http://www.vision.caltech.edu/Image_Datasets/Caltech256/ http://cocodataset.org/#home https://ai.stanford.edu/~jkrause/cars/car_dataset.html https://www.kaggle.com/c/dogs-vs-cats http://www.ifp.illinois.edu/~vuongle2/helen/ https://www.kaggle.com/moltean/fruits/kernels

I hope this helps,

Jan