Photometric stereo training data set with annotated leaf masks

Abstract

Automated leaf segmentation is a challenging area in computer vision. Recent advances in machine learning approaches allowed to achieve better results than traditional image processing techniques; however, training such systems often require large annotated data sets. To contribute with annotated data sets and help to overcome this bottleneck in plant phenotyping research, here we provide a novel photometric stereo (PS) data set with annotated leaf masks. This data set forms part of work done in the BBSRC Tools and Resources Development project BB/N02334X/1.