Recently, the crop quality intelligent perception team of Hefei Institute of Intelligent Machinery, Hefei Institutes of Physical Science (HFIPS) of Chinese Academy of Sciences (CAS) has developed a new algorithm for near infrared (NIR) spectroscopy, suitable for high-throughput identification of the authenticity of crop varieties.
The authenticity of crop varieties is of great significance in variety protection and seed breeding. Traditional methods for authentic identification of crop varieties, such as DNA molecular identification, isoenzyme identification and field identification, have the disadvantages of complicated operations, time-consuming, sample damage, environmental pollution and slow detection results, so an effective method is urgently needed to realise the authenticity identification of crop varieties.
As a rapid detection technology, NIR spectroscopy, has many advantages. It is environmentally friendly, sensitive and non-destructive. In this research, a self-developed high-throughput seed quality sorting instrument based on NIR spectroscopy, made by team, can achieve rapid identification and sorting of individual seeds. Based on this instrument, researchers proposed an improved convolution neural network (CNN)—the InResSpectra network—to help achieve more accurate seed variety identification. This is an optimised Inception network and successfully removed the 1 × 1 convolution branch to reduce the complexity of the model, and increased the residual element of the ResNet network, which accelerated the training of the neural network and improved accuracy.
Researchers applied the developed system and the InResSpectra network for the identification of 24 wheat varieties and 21 rice varieties, and achieved 95.35 % and 93.07 % accuracy, respectively, which provided an effective method for the spectroscopic identification of the authenticity of crop varieties.