Early detection of some crops responses to water stress using imaging and artificial intelligence techniques

الملخص

The current study was done to evaluate the efficiency of imaging techniques, determine the accuracy and reliability of artificial intelligence models in the early detection of various types of environmental stresses, and study the possibility of combining imaging techniques and artificial intelligence into an integrated system that can be used in agricultural practices, particularly for wheat (Triticum aestivum L.) and oat (Avena sativa L.) crops. The plants were divided into four groups according to environmental stress factors. The control group included unstressed plants; the second group included plants exposed to water stress (drought). The results of the visual spectrum analysis of water stress images showed that the yellow color intensity of wheat plants was measured at 12.62% on day (10) and a burn rate of 9.71% for the same day. In oat plants, the intensity was 0.91% on day (10) and a burn rate of 10.82% for both plants. The results of the thermal image analysis of water stress in both plants showed an increase in the intensity of red color and the development of leaf reddening. This intensity began to appear in the image (F.bmp) at 0.93% and reached its maximum in the image (J.bmp), recording a color intensity of 10.58% in wheat plants. In oat plants, the intensity began to appear in the image (F.bmp), recording 3.17%, and reached its maximum in the image (J.bmp), recording 10.82%, and indicating deterioration in the physiological condition of the plants. Based on the above, oat plants showed greater sensitivity to the three types of stresses compared to thermal imaging. The study demonstrated that thermal imaging technology was more effective in early detection of stresses in both plants. The results of combining imaging and artificial intelligence techniques using PlantDoc AI and Plantix for both plants and all stresses also showed that both programs provided confidence levels ranging from 70-96% for wheat plants and 60-97% for oat plants, enhancing the accuracy of early detection of stresses to which plants are exposed.