This code was developed for 2018 data science bowl competition for automation of cell nucleus detection. This version of mask RCNN was developed by Coco for object segmentation. The dataset is available on kaggle website, or could be downloaded from github. The human annotated dataset contains a large number of segmented nuclei images. This competition was evaluated on the mean average precision at different intersection over union (IoU) thresholds. The submission was prepared with run length encoding of cell nucleus pixels.