Unmanned aerial vehicles (UAVs) are being increasingly viewed as valuable tools to aid the management of farms. This kind of technology can be particularly useful in the context of extensive cattle farming, as production areas tend to be expansive and animals tend to be more loosely monitored. With the advent of deep learning, and convolutional neural networks (CNNs) in particular, extracting relevant information from aerial images has become more effective. Despite the technological advancements in drone, imaging and machine learning technologies, the application of UAVs for cattle monitoring is far from being thoroughly studied, with many research gaps still remaining. In this context, the objectives of this study were threefold: (1) to determine the highest possible accuracy that could be achieved in the detection of animals of the Canchim breed, which is visually similar to the Nelore breed (Bos taurus indicus); (2) to determine the ideal ground sample distance (GSD) for animal detection; (3) to determine the most accurate CNN architecture for this specific problem. The experiments involved 1853 images containing 8629 samples of animals, and 15 different CNN architectures were tested. A total of 900 models were trained (15 CNN architectures × 3 spacial resolutions × 2 datasets × 10-fold cross validation), allowing for a deep analysis of the several aspects that impact the detection of cattle using aerial images captured using UAVs. Results revealed that many CNN architectures are robust enough to reliably detect animals in aerial images even under far from ideal conditions, indicating the viability of using UAVs for cattle monitoring.

The management of livestock in extensive production systems may be challenging, especially in large areas. Using Unmanned Aerial Vehicles (UAVs) to collect images from the area of interest is quickly becoming a viable alternative, but suitable algorithms for extraction of relevant information from the images are still rare. This article proposes a method for counting cattle which combines a deep learning model for rough animal location, color space manipulation to increase contrast between animals and background, mathematical morphology to isolate the animals and infer the number of individuals in clustered groups, and image matching to take into account image overlap. Using Nelore and Canchim breeds as a case study, the proposed approach yields accuracies over 90% under a wide variety of conditions and backgrounds.

Crop models can aid the synthesis and application of knowledge, planning of experiments and forecasting in agricultural systems. Few studies have reviewed the uses and applications of these models for tropical forages. The purpose of this study was to review the information available in this scientific area, highlighting the main models, their applications and limitations. Several empirical models have been developed to predict the growth and biomass accumulation of tropical forages, especially for the genera Cynodon, Paspalum, Panicum and Brachiaria. Their application, however, is often location or region specific. The adaptation of mechanistic models to accurately predict biomass accumulation in tropical grasses is still limited. Recent advances have been made on the plot‐scale and farm‐scale process‐based models ALMANAC, CROPGRO Perennial Forage and agricultural production systems simulator (APSIM), with promising results. In addition, global‐scale process‐based models, such as the Century Agroecosystem Model and the Orchidee Grassland Management Model, have been tested for tropical grassland areas. A greater number of region‐specific calibrations of empirical models can enhance their use, and improved databases and model parameterizations for a wide range of tropical grasses will enable the continuous improvement of mechanistic models.

Unmanned Aerial Vehicles (UAVs) are being increasingly viewed as valuable tools to aid the management of farms. This kind of technology can be particularly useful in the context of extensive cattle farming, as production areas tend to be expansive and animals tend to be more loosely monitored. With the advent of deep learning, and Convolutional Neural Networks (CNNs) in particular, extracting relevant information from aerial images has become more effective. Despite the technological advancements in drone, imaging and machine learning technologies, the application of UAVs for cattle monitoring is far from being thoroughly studied, with many research gaps still remaining. In this context, the objectives of this study were threefold: 1) to determine the highest possible accuracy that could be achieved in the detection of animals of the Canchim breed, which is visually similar to the Nelore breed (\textit{Bos taurus indicus}); 2) to determine the ideal Ground Sample Distance (GSD) for animal detection; 3) to determine the most accurate CNN architecture for this specific problem. The experiments involved 1,853 images containing 8,629 samples of animals, and 15 different CNN architectures were tested. A total of 900 models were trained (15 CNN architectures * 3 spacial resolutions * 2 datasets * 10-fold cross validation), allowing for a deep analysis of the several aspects that impact the detection of cattle using aerial images captured using UAVs. Results revealed that many CNN architectures are robust enough to reliably detect animals in aerial images even under far from ideal conditions, indicating the viability of using UAVs for cattle monitoring.

The evolution in imaging technologies and artificial intelligence algorithms, coupled with improvements in UAV technology, has enabled the use of unmanned aircraft in a wide range of applications. The feasibility of this kind of approach for cattle monitoring has been demonstrated by several studies, but practical use is still challenging due to the particular characteristics of this application, such as the need to track mobile targets and the extensive areas that need to be covered in most cases. The objective of this study was to investigate the feasibility of using a tilted angle to increase the area covered by each image. Deep Convolutional Neural Networks (Xception architecture) were used to generate the models for animal detection. Three experiments were carried out: (1) five different sizes for the input images were tested to determine which yields the highest accuracies; (2) detection accuracies were calculated for different distances between animals and sensor, in order to determine how distance influences detectability; and (3) animals that were completely missed by the detection process were individually identified and the cause for those errors were determined, revealing some potential topics for further research. Experimental results indicate that oblique images can be successfully used under certain conditions, but some practical limitations need to be addressed in order to make this approach appealing.

O objetivo deste trabalho foi avaliar a produtividade e as características biométricas do capim-braquiária (Urochloa decumbens) sob pastejo, em sistema silvipastoril. Avaliou-se um sistema com árvores nativas de interesse comercial, plantadas em renques, no sentido norte-sul, espaçados por 17 m. Utilizou-se o delineamento experimental de blocos ao acaso, com quatro repetições e parcelas subdivididas, tendo-se avaliado o ciclo de rebrota da forragem nas parcelas, e a distância desta em relação aos renques, nas subparcelas. Sete ciclos de rebrota foram avaliados, em quatro distâncias dos renques: a 2 e a 6 m do renque leste, e a 2 e a 6 m do renque oeste. Quantificou-se a produtividade de matéria seca, o índice de área foliar (IAF), a área foliar específica (AFE) e a altura das plantas. A produtividade de forragem, nos dois primeiros ciclos, e o IAF, no segundo e no terceiro ciclo, diminuiram com níveis de sombra acima de 39 e 40%, respectivamente. Já a altura de plantas aumentou com sombreamentos superiores a 53%, nos quatro primeiros ciclos; o que também ocorreu com a AFE, com sombreamento superior a 66%, nos três primeiros ciclos.