some issues about dataset
monkilee opened this issue · comments
Thanks for providing such a great dataset. Regarding to the dataset, there may be some problem extracting the dataset from zip. It should contain 300000 images, am I right? But I've got less than 300000 images. When I tried to reconver the broken files from gzrecover, It still has less than 300000 images. And Also there is no image paths in AI Challenge annotation files that map the annotation data to the ground-truth images. I tried to map the annotation in order but it seem some files are over 300000 indexes in image names. Could you provide a some guideline to parse the ai challenge datasets?
Thank you for your suggestion. We will check the uploaded datasets and provide a guideline for the users.
Thanks for providing such a great dataset. Regarding to the dataset, there may be some problem extracting the dataset from zip. It should contain 300000 images, am I right? But I've got less than 300000 images. When I tried to reconver the broken files from gzrecover, It still has less than 300000 images. And Also there is no image paths in AI Challenge annotation files that map the annotation data to the ground-truth images. I tried to map the annotation in order but it seem some files are over 300000 indexes in image names. Could you provide a some guideline to parse the ai challenge datasets?