Comparing CNNs and PLSr for estimating wheat organs biophysical variables using proximal sensing

Estimation of biophysical vegetation variables is of interest for diverse applications, such as monitoring of crop growth and health or yield prediction. However, remote estimation of these variables remains challenging due to the inherent complexity of plant architecture, biology and surrounding environment, and the need for features engineering. Recent advancements in deep learning, particularly convolutional neural networks (CNN), offer promising solutions to address this challenge. Unfortunately, the limited availability of labeled data has hindered the exploration of CNNs for regression tasks, especially in the frame of crop phenotyping. In this study, the effectiveness of various CNN models in predicting wheat dry matter, nitrogen uptake, and nitrogen concentration from RGB and multispectral images taken from tillering to maturity was examined. To overcome the scarcity of labeled data, a training pipeline was devised. This pipeline involves transfer learning, pseudo-labeling of unlabeled data and temporal relationship correction. The results demonstrated that CNN models significantly benefit from the pseudolabeling method, while the machine learning approach employing a PLSr did not show comparable performance. Among the models evaluated, EfficientNetB4 achieved the highest accuracy for predicting above-ground biomass, with an R² value of 0.92. In contrast, Resnet50 demonstrated superior performance in predicting LAI, nitrogen uptake, and nitrogen concentration, with R² values of 0.82, 0.73, and 0.80, respectively. Moreover, the study explored multi-output models to predict the distribution of dry matter and nitrogen uptake between stem, inferior leaves, flag leaf, and ear. The findings indicate that CNNs hold promise as accessible and promising tools for phenotyping quantitative biophysical variables of crops. However, further research is required to harness their full potential.

Quantitative macromolecular patterns in phytoplankton communities resolved at the taxonomical level by single-cell Synchrotron FTIR-spectroscopy

BACKGROUND

Technical limitations regarding bulk analysis of phytoplankton biomass limit our comprehension of carbon fluxes in natural populations and, therefore, of carbon, nutrients and energy cycling in aquatic ecosystems. In this study, we took advantage of Synchrotron FTIR micro-spectroscopy and the partial least square regression (PLSr) algorithm to simultaneously quantify the protein, lipid and carbohydrate content at the single-cell level in a mock phytoplankton community (composed by a cyanobacterium, a green-alga and a diatom) grown at two temperatures (15 °C and 25 °C).

RESULTS

The PLSr models generated to quantify cell macromolecules presented high quality fit (R² ≥ 0.90) and low error of prediction (RMSEP 2-6% of dry weight). The regression coefficients revealed that the prediction of each macromolecule was not exclusively dependent on spectral features corresponding to that compound, but rather on all major macromolecular pools, reflecting adjustments in the overall cell carbon balance. The single-cell analysis, studied by means of Kernel density estimators, showed that the modes of density distribution of macromolecules were different at 15 °C and 25 °C. However, a substantial proportion of cells was biochemically identical at the two temperatures because of population heterogeneity.

CONCLUSIONS

The spectroscopic approach presented in this study allows the quantification of macromolecules in single phytoplankton cells. This method showed that population heterogeneity most likely ensures a backup of non-acclimated cells that may rapidly exploit new favourable niches. This finding may have important consequences for the ecology of phytoplankton populations and shows that the "average cell" concept might substantially limit our comprehension of population dynamics and biogeochemical cycles in aquatic ecosystems.