LoannPeurey
commented
1 month ago Idea of having a 'sets.csv' file listing sets is interesting, it would hold information aggregated by sets like segmentation, origin, notes etc, the package should create it if does not exist but info should be added there
Is your feature request related to a problem? Please describe. It is not always clear what annotation set is human vs machine, especially in the segmentation part. Because although some annotation sets are labeled to be recognizable between e.g. VTC and eaf, some human annotation are based on machine segmentations (human classifies the segments vtc identified)
Describe the solution you'd like New field in annotations.csv index . Being 'segmentation' and having to be between human or machine