Thank you for uploading the exact benchmark datasets that you used to validate Img2Mol!
I have a question regarding the size of the different sets. It appears that you have removed some images from the original benchmark sets. For example, the JPO dataset normally consists of 450 images but here it is 365. Could you share the criteria for the removal of images from the sets?
Thanks in advance! :)
Thank you for uploading the exact benchmark datasets that you used to validate Img2Mol! I have a question regarding the size of the different sets. It appears that you have removed some images from the original benchmark sets. For example, the JPO dataset normally consists of 450 images but here it is 365. Could you share the criteria for the removal of images from the sets? Thanks in advance! :)