Open tanpengjie opened 3 months ago
Dear Pengjie,
Thank you for your interest in my dataset. The dataset was designed primarily for anomaly detection purposes rather than localization, which is why ground truth labels for the type 7-10 datasets are missing. Moreover, the available labels for the other datasets are not accurate (or even just black images), as I do not have the actual segmentation maps.
Unfortunately, I currently do not have the time to generate these ground truth labels myself. However, I encourage you to contribute by creating these labels. Should you be able to generate the ground truth data, either manually or using a state-of-the-art unsupervised anomaly detection and localization method, I would be delighted to incorporate your contributions into the dataset.
Thank you once again for your interest and understanding.
Best wishes,
Dear Thomine
Well noted. Thanks for your contribution to this field again. In the next period of time, I will spend some time labeling the rest of the data and checking the labeled data.
Thanks.
TAN Pengjie
Thanks a lot! Please let me know once you've finished labeling the remaining data. Appreciate your efforts.
Dear Thomine
Thanks for your contribution to the unsupervised defect detection and localization field. I am interested in your open-sourced dataset. However, I find that there is no corresponding label for the type 7, 8,9,10 datasets. If you are free, could you release them? Thanks for your contributions again.
Best Wishes
TAN Pengjie