CT-OCR-2022 (scroll003)

CT-OCR-2022 dataset

CT-OCR-2022 dataset contains optically scanned images for source paper document, 400 X-ray projections, 2687 CT-reconstructed cross-sections and  segmentation markups for 6 model objects.

Description of the data for each model object is presented in the table.

            Files
            Data description

            [package].proj_src/proj_s_.log
            X-ray measurement log file

            [package].proj_src/*.tif
            preprocessed projections before rotation axe correction

            [package].proj_norm/proj_s_.log
            X-ray measurement and geometry correction log file

            [package].proj_norm/*.tif
            preprocessed projections after rotation axe correction

            [package].rec_XXXX/metadata.json
            reconstruction metadata

            [package].rec_XXXX/*.tif
            CT-reconstructed volume, slices with size XXXX×XXXX

            [package].seg_XXXX/*.tif.seg.png
            segmentation markup

            [package].blank.png
            sample croped from pdf

            [package].scan.png
            sample croped from scanned image

Due to the large amount of data, folders were packed into multi-volume zip-archives. Dataset published in Zenodo service in several linked repositories.

Any questions, complaints, etc. can be directed to: [e-mail] (Dmitry Polevoy)

The article about this dataset is currently under review. When the article is published, the full reference will be added.