Related Experiment Video
Updated: Nov 9, 2025
11:46
High and Low Throughput Screens with Root-knot Nematodes Meloidogyne spp.
Published on: March 12, 2012
28.4K
SeedQuant: a deep learning-based tool for assessing stimulant and inhibitor activity on root parasitic seeds
Justine Braguy1,2, Merey Ramazanova3, Silvio Giancola3
1Division of Biological and Environmental Science and Engineering, the BioActives Lab, King Abdullah University of Science and Technology, Thuwal 23955-6900, Saudi Arabia.
Plant Physiology
|April 15, 2021
Summary
An automated tool, SeedQuant, uses deep learning to accurately count germinated parasitic plant seeds (Striga spp. and Orobanchaceae) in bioassays. This significantly speeds up research for crop protection strategies against these damaging weeds.
Area of Science:
- Agricultural Science
- Computational Biology
- Plant Science
Background:
- Root parasitic plants like witchweeds (Striga spp.) and broomrapes (Orobanchaceae, Phelipanche spp.) cause significant crop yield losses globally.
- Their obligate parasitic nature necessitates host-derived stimulants for seed germination, presenting a target for control strategies.
- Current in vitro germination assays are sensitive but labor-intensive, hindering high-throughput screening.
Purpose of the Study:
- To develop an automated method for counting germinated parasitic seeds in bioassays.
- To accelerate the screening process for germination stimulants and inhibitors for Striga and Orobanchaceae species.
- To improve the efficiency and accuracy of parasitic plant seed germination assessment.
Main Methods:
- Utilized deep learning, specifically the Faster Region-based Convolutional Neural Network (Faster R-CNN) algorithm, for object detection.
- Developed an automatic seed census tool, SeedQuant, to discriminate between germinated seeds (GS) and non-germinated seeds.
- Applied the method to count Striga hermonthica seeds in bioassay images.
Main Results:
- The SeedQuant tool achieved 94% accuracy in counting Striga hermonthica seeds.
- The automated counting process reduced the time per image from approximately 5 minutes to 5 seconds.
- Demonstrated the potential for accelerating parasitic seed germination bioassays.
Conclusions:
- SeedQuant offers a highly accurate and efficient solution for assessing parasitic plant seed germination.
- The open-source software facilitates high-throughput screening, aiding the development of novel crop protection methods.
- SeedQuant can be further trained for diverse seed counting applications in research.
