Missing Images in WIDER Dataset:

[yujun686]

Hello, the WIDER dataset downloaded from your provided link contains 57,442 images, but there are 57,491 labels. During training, some images cannot be located due to path errors. How can this issue be resolved?

[NOP1224]

The number of images in the processed WIDER dataset we provided is 57,491 instead of 57,442. it is recommended that you re-check the dataset

[yujun686]

I re-downloaded the split_image.zip from the WIDER dataset via the Google Drive link on the homepage, but after extraction, it still contains only 57,442 files. Could you please verify if the link is correct? Thank you very much!

[NOP1224]

Sorry, this was an oversight on our part and we will re-upload the WIDER dataset.