This dataset includes 16 images of cell culture flasks containing fixed and stained cell colonies used for a colony formation assay of the T-47D (breast) cancer cell line. Each flask contains cell colonies, as well as background structures (e.g. shadows) and outer contours of the cell flask. These images were obtained from a flatbed laser scanner (Epson Perfection V850 Pro), providing rgb images with a resolution of 2125\(\times\)2985, 1200 dots per inch (dpi), 21.17 \(\mu\)m/pixel spatial resolution and 48-bit depth. No prior filtering nor adjustments were performed on the captured images during scanning with the scanner software (EPSON Scan v3.9.3.3). The goal of this This dataset was used for identification, segmentation and counting of viable colonies (conglomerations composed of > 50 cells) by means of a developed segmentation algorithm. The algorithm was benchmarked against 3 trained, independent human observers and one extra independent observer establishing a ground truth (GT) by manual counting during a microscopic analysis. In particular, each observer presented detected colonies with x- and y-coordinates (given in respective columns in each .txt file).