Counting Pollen Viability via Deep Learning

Pollen counts are an important measure for biologists working in plant phenotyping and assist in understanding a plant’s ability to tolerate stresses resulting from different growing conditions. Accurately assessing pollen can prove challenging within high-throughput experiments [2]. Current pollen counting methods include: manual counting, image analysis software [1], & impedance flow cytometry [4]. However, these methods are cumbersome, don’t provide viability (e.g. fertile vs. sterile pollen grains) counts, and are costly to implement, respectively.