Hey Luca,
I have encountered problems to prepare the dataset. As the information from the README file, the original dataset could be downloaded directly from COCO or kaggle. However, when I am tracing into the split_data.py file, I found the lines 5 and 6 are the original input of the train_validate and test npy files. I wanna ask you here is that how did you prepare these npy files. Is that directly download from the COCO? or did you do additional preprocessing after getting the original train, valid and test files from the COCO website?
Best Regards,
Peng
Hey Luca,
I have encountered problems to prepare the dataset. As the information from the README file, the original dataset could be downloaded directly from COCO or kaggle. However, when I am tracing into the split_data.py file, I found the lines 5 and 6 are the original input of the train_validate and test npy files. I wanna ask you here is that how did you prepare these npy files. Is that directly download from the COCO? or did you do additional preprocessing after getting the original train, valid and test files from the COCO website?
Best Regards,
Peng