Automated seed germination measurements

For NIOO-KNAW, we developed a computer vision system that automatically measures the seed germination rate of parasitic weeds. They use this system to automatically assess the effectiveness of their biological treatments against parasitic weeds, aiming to prevent failed harvests in Africa.

The automated seed germination measurements are based on deep learning. We trained an algorithm to recognize seeds, seeds that are germinated, and to ignore any debris or other impurities. The output is instant and tells the researchers directly what they need to know.

Before this system was in use, highly educated researchers were counting the (germinated) seeds by hand. This is not only a tedious and expensive task, it was also causing a bottleneck in the whole research. Our computer vision system takes away this bottleneck. The researchers can now move to high-throughput experiments with more repetitions and variations.

See our Detecting Varroa mite in honey bee combs project for more work that we do for researchers.