I was looking at the Dark Face dataset available on https:\/\/flyywh.github.io\/CVPRW2019LowLight\/<\/a> . Instead, all I got was: Now I may be missing something very obvious, but I have searched online quite a bit but I can’t figure out how to get the complete dataset. They said it was part of competion ‘UG 2+ challenge’. website: http:\/\/cvpr2023.ug2challenge.org\/<\/a><\/p>\n Can someone please help me with link(s) to where I can access the dataset which is properly structure.<\/p>\n<\/div>\n submitted by \/u\/Altruistic-Hat-9604 <\/a>
About the dataset according to the description of the dataset:
DARK FACE dataset provides 6,000 real-world low light images<\/strong> captured during the nighttime, at teaching buildings, streets, bridges, overpasses, parks etc., all labeled with bounding boxes for of human face, as the main training and\/or validation sets. We also provide 9,000 unlabeled low-light images<\/strong> collected from the same setting. Additionally, we provided a unique set of 789 paired low-light\/normal-light images<\/strong> captured in controllable real lighting conditions (but unnecessarily containing faces), which can be used as parts of the training data at the participants’ discretization. There will be a hold-out testing set of 4,000 low-light images<\/strong>, with human face bounding boxes annotated.
When I downloaded the dataset, I was hoping for the folders to be structured something like:
Dark Face Dataset
\u251c\u2500\u2500 Low-light
\u2502 \u251c\u2500\u2500 Training
\u2502 \u2502 \u251c\u2500\u2500 Positive
\u2502 \u2502 \u2514\u2500\u2500 Negative
\u2502 \u2514\u2500\u2500 Validation
\u2502 \u251c\u2500\u2500 Positive
\u2502 \u2514\u2500\u2500 Negative
\u251c\u2500\u2500 Paired Low-light and Normal-light
\u2502 \u251c\u2500\u2500 Training
\u2502 \u2514\u2500\u2500 Validation
\u2514\u2500\u2500 Testing
\u2514\u2500\u2500 Low-light<\/p>\n
images: containing 6000 images
labels: containing 6000 text files containg information about images. But nowhere is the mention of training,validation, no low-light and normal-light images. <\/p>\n
[link]<\/a><\/span> [comments]<\/a><\/span><\/p>
