A Synthetic Dataset for Stirrup Rebar Segmentation
The dataset contains:
A synthetic training set of 12,000 images, and a synthetic validation set of 4,000.
A synthetic test set of 4,800 images (only top rebars are annotated).
A real-world test set of 233 images (only top rebars are annotated).
Diverse rebar specifications, stacking, lighting, distractors, and background conditions.
mkdir -p train_syn/train2017
mv train_syn/train2017_sub{1,2,3}/*… See the full description on the dataset page:
https://huggingface.co/datasets/tsrobcvai/Synthetic_Dataset_for_Stirrup_Rebar_Segmentation.
