Hi, I have a dataset that is converted from videos into images with annotations in KITTI format. I’m using this dataset for object detection. Unfortunately, the script that converted the videos into images seems to have corrupted a few files and there are a few images that don’t have corresponding label files. It is quite time consuming to run the dataset in DIGITS for several hours only to have DIGITS abort on an file mismatch or an file that can’t be interpreted. Is there a flag or setting I can use to skip files that have errors and just continue with creating the dataset?