Non-Destructive Testing of Moisture and Nitrogen Content in Pinus Massoniana Seedling Leaves with NIRS Based on MS-SC-CNN

Abstract

Pinus massoniana is a pioneer reforestation tree species in China. It is crucial to evaluate the seedling vigor of Pinus massoniana for reforestation work, and leaf moisture and nitrogen content are key factors used to achieve it. In this paper, we proposed a non-destructive testing method based on the multi-scale short cut convolutional neural network (MS-SC-CNN) to measure moisture and nitrogen content in leaves of Pinus massoniana seedlings. By designing a reasonable short cut structure, the method realized the transmission of loss function gradient across the multi-layer structure in the network and reduced the information loss caused by the multi-layer transmission in the forward propagation. Meanwhile, in the back propagation stage, the loss caused by the multi-layer transmission of gradient was reduced. Thus, the gradient vanishing problem in training was avoided. Since the method realized cross-layer transmission error, the convolutional layer could be increased appropriately to obtain higher measurement accuracy. To verify the performance of the proposed MS-SC-CNN non-destructive measurement method, the near-infrared hyperspectral data of sample leaves of 219 Pinus massoniana seedlings were collected from the Huangping Forest Farm in Guizhou Province. The correlation coefficient between the measured and real values of the prediction was as high as 0.977 and the root mean square error was 0.242 for the moisture content of Pinus massoniana seedling leaves. For the nitrogen content of Pinus massoniana seedling leaves, the correlation coefficient between the measured and real values of the prediction was 0.906 and the root-mean-square error was 0.061. The results showed that the non-destructive testing method based on MS-SC-CNN that we proposed can accurately measure the moisture and nitrogen content in leaves of Pinus massoniana seedlings.

Moisture content prediction of semen ziziphi spinosae based on hyperspectral images coupled with convolutional neural networks and subregional voting

“…The existing studies lack the explanation of the model mechanism to understand the feature learning process by CNN in spectral analysis 31,32 . The feature extraction process in this study exactly fills this gap.…”

“…The existing studies lack the explanation of the model mechanism to understand the feature learning process by CNN in spectral analysis. 31,32 The feature extraction process in this study exactly fills this gap. This study proposes that the CNN model has three convolutional layers, and plotting the features of the convolutional layers shows how the spectrum changes in different convolutional layers.…”

Identification of Geographical Origin of Chinese Chestnuts Using Hyperspectral Imaging with 1D-CNN Algorithm

“…Normalization was performed to facilitate comparison and weighting between different spectral variables. Derivatives (1der and 2der) were implemented to eliminate baseline offsets based on a 5-point Savitzky-Golay algorithm within quadratic function convolution [43]. All the above data preprocessing algorithms were implemented in Unscramble X10.4 (CAMO Software, Oslo, Norway).…”

3D Face Recognition Based on an Attention Mechanism and Sparse Loss Function

“…Therefore, the attention mechanism coupled with the convolutional layers was constructed as the backbone network to focus on the key information of the preprocessed data in order to reduce the influence of different facial expressions on recognition accuracy. The convolutional layer is known as a grid-like topological specialized kind of neural network which has gained tremendous success in practical applications [34,35]. This kind of neural network layer is able to extract robust features from images and is utilized as a feature extractor in this study.…”