Abstract:

China is a major agricultural country. With the rapid development of agricultural science and technology, agriculture has entered a new stage of development with high yield, high quality, and high efficiency. Achieving automation and intelligent observation of crop growth period recognition is a crucial part of agricultural modernization. This paper introduced the current research status of crop growth period recognition and presented two methods for automatic observation and identification of winter wheat growth period, one based on the Normalized Difference Vegetation Index (NDVI) and the other based on deep learning. Using winter wheat in Henan as an example, the results of automatic observation and identification from both methods were compared with manual observations. The results validated the feasibility and effectiveness of both identification methods, showing high accuracy and efficiency, thereby improving measurement efficiency and reliability. In terms of identification accuracy, the two methods had their own strengths at different growth periods and could complement each other. The deep learning-based identification method demonstrated better generalizability compared to the NDVI-based method. However, both methods required optimization and upgrading in the future to further enhance identification accuracy.

Key words: winter wheat, growth period, image recognition, normalized differential vegetation index (NDVI), deep learning