New agricultural practices and land-use intensification in the Cerrado biome have affected the soil carbon stocks. A major part of the native vegetation of the Brazilian Cerrado, a tropical savanna-like ecoregion, has been replaced by crops, which has caused changes in the soil carbon (C) stocks. To ensure the sustainability of this intensified agricultural production, actions have been taken to increase soil C stocks and mitigate greenhouse gas emissions. In the last two decades, new agricultural practices have been adopted in the Cerrado region, and their impact on C stocks needs to be better understood. This subject has been addressed in a systematic review of the existing data in the literature, consisting of 63 articles from the Scopus database. Our review showed that the replacement of Cerrado vegetation by crop species decreased the original soil C stocks (depth 0–30 cm) by 73%, with a peak loss of 61.14 Mg ha⁻¹. However, when analyzing the 0–100 cm layer, 52.4% of the C stock data were higher under cultivated areas than in native Cerrado soils, with a peak gain of 93.6 Mg ha⁻¹. The agricultural practices implemented in the Brazilian Cerrado make low-carbon agriculture in this biome possible.

In image classification, few-shot learning deals with recognizing visual categories from a few tagged examples. The degree of expressiveness of the encoded features in this scenario is a crucial question that needs to be addressed in the models being trained. Recent approaches have achieved encouraging results in improving few-shot models in deep learning, but designing a competitive and simple architecture is challenging, especially considering its requirement in many practical applications. This work proposes an improved few-shot model based on a multi-layer feature fusion (FMLF) method. The presented approach includes extended feature extraction and fusion mechanisms in the Convolutional Neural Network (CNN) backbone, as well as an effective metric to compute the divergences in the end. In order to evaluate the proposed method, a challenging visual classification problem, maize crop insect classification with specific pests and beneficial categories, is addressed, serving both as a test of our model and as a means to propose a novel dataset. Experiments were carried out to compare the results with ResNet50, VGG16, and MobileNetv2, used as feature extraction backbones, and the FMLF method demonstrated higher accuracy with fewer parameters. The proposed FMLF method improved accuracy scores by up to 3.62% in one-shot and 2.82% in five-shot classification tasks compared to a traditional backbone, which uses only global image features.