Sign in Subscribe
Agricultural industry

Deep Learning Assisted Nitrogen Stress Detection for Improved Crop Yields

Strategic Insights

— 1 min read

Researchers at the Indian Council of Agricultural Research (ICAR) have developed a Variable Rate fertilizer Application (VRA) system assisted by a Deep Learning (DL) model to detect nitrogen stress in wheat crops and optimize fertilizer application. The system uses embedded technology and RGB image analysis to classify nitrogen stress in real-time, achieving high precision and recall rates. The study reports a significant reduction in nitrogen fertilizer usage, resulting in cost savings and improved crop yields.

Key Takeaways:

  • The VRA system, assisted by the AlexNet DL model, achieved precision, recall, and F1-score of 0.977, 0.973, and 0.973, respectively, for classifying nitrogen stress into three classes.
  • The system can operate at an operational speed of 0.4 m/s with a field capacity of 0.32 ha/h in a 26 DAS wheat crop.
  • The VRA system resulted in a consistent difference in vegetation indices (ExG, RGRI, VARI, and NGRDI) before and after operation, indicating uniformity of operation throughout the field.
  • The average percentage nitrogen fertilizer saving under VRA as compared to traditional technique was 37.53%.
  • The study demonstrates the effectiveness of the VRA system in real-time nitrogen stress detection and its potential to improve crop yields and reduce fertilizer usage.

Statistics:

  • 0.977: precision achieved by the AlexNet DL model for classifying nitrogen stress.
  • 0.973: recall achieved by the AlexNet DL model for classifying nitrogen stress.
  • 0.973: F1-score achieved by the AlexNet DL model for classifying nitrogen stress.
  • 0.4 m/s: operational speed of the VRA system.
  • 0.32 ha/h: field capacity of the VRA system.
  • 26 days after sowing (DAS): wheat crop duration during which the VRA system was tested.
  • 37.53%: average percentage nitrogen fertilizer saving under VRA as compared to traditional technique.

Sources:

  • Deep Learning Assisted Real-time Nitrogen Stress Detection for Variable Rate Fertilizer Applicator In Wheat Crop. Computers and Electronics In Agriculture, 2025; 237.
  • Indian Council of Agricultural Research (ICAR). Journal of Engineering. October 13, 2025; p 771.

Read more