Difficulties when trying to reproduce certain examples from the paper. #16
Description
Hello, authors!
Thanks for sharing this awesome work! I was giving a shot at reproducing some of the experimental results shown in the paper using code from the single-image notebook shared here. Altering the gsd values works fine for me but when I tried to either a) alter a stadium's appearance by only altering the values of the coordinate pairs, or b) alter the appearance of an example either belonging to the class "crop_field" or the class "park" by altering the value of the "month" metadatum (expecting, thus, diminished flora during the winter months and vice-versa) any changes in the resulting image where close to being negligible. I am possibly doing something wrong but would it be possible for you to look into this and, maybe, share any insights into how I could safely reproduce the reported results? Thank you in advance!