In recent years, deep learning models have become the standard for agricultural computer vision. Such models are typically fine-tuned to agricultural tasks using model weights that were originally fit to more general, non-agricultural datasets. This lack of agriculture-specific fine-tuning potentially increases training time and resource use, and decreases model performance, leading an overall decrease in data efficiency. To overcome this limitation, we collect a wide range of existing public datasets for three distinct tasks, standardize them, and construct standard training and evaluation pipelines, providing us with a set of benchmarks and pretrained models. We then conduct a number of experiments using methods which are commonly used in deep learning tasks, but unexplored in their domain-specific applications for agriculture. Our experiments guide us in developing a number of approaches to improve data efficiency when training agricultural deep learning models, without large-scale modifications to existing pipelines. Our results demonstrate that even slight training modifications, such as using agricultural pretrained model weights, or adopting specific spatial augmentations into data processing pipelines, can significantly boost model performance and result in shorter convergence time, saving training resources. Furthermore, we find that even models trained on low-quality annotations can produce comparable levels of performance to their high-quality equivalents, suggesting that datasets with poor annotations can still be used for training, expanding the pool of currently available datasets. Our methods are broadly applicable throughout agricultural deep learning, and present high potential for significant data efficiency improvements.

The kallikrein kinin system (KKS) is involved in blood pressure (BP) regulation by a mechanism not completely defined. Our previous reports have shown that high K + intake or prepuberal gonadectomy (Gx) diminish BP with a simultaneous increase in urine kallikrein activity (UKa) and plasma aldosterone (PA) levels, revealing a link between those systems. Thus, since K + may be involved in the regulation of KKS, we explored the rectifying outer medulla K + (ROMK) channel blockade using glibenclamide (Gli) in different gonad contexts.Spontaneously hypertensive rats of both sexes, half of them Gx at weaning, were studied at 12 weeks of age (n = 32). Glucose solution (4%) with or without Gli (10 mg/kg bwt) was orally administered in the last 3 days of the experiment. We analyzed BP, glomerular filtration rate, PA, daily urine Na + and K + excretion and UKa. Renal cortex kallikrein activity (RKa) and UKa were determined by colorimetric assay. Renal mRNA levels of Kcjn1 (ROMK), Atp1α1 (Na + K + Atpase) and Klk1 (kallikrein 1) genes were determined by quantitative real time PCR. Urine Na + /K + increased after Gli treatment (0.55 ± 0.03 vs 1.34 ± 0.30, p <0.05) due to a K + excretion decrease in intact male and ovariectomized rats and to a Na + excretion increase in intact female. These changes were concomitant with increased GFR within the normal range (0.51 ± 0.06 vs 0.76 ± 0.06 ml/min/100g bwt p < 0.01) and no differences in BP and PA among groups. After Gli treatment, renal cortex Klk1 and RKa levels increased in intact males (297 and 137 %, p < 0.05, respectively), while in orchidectomized group Klk1 and UKa levels increased (179 and 230 %, p < 0.05; respectively). Kcjn1 and Atp1α1 mRNA levels decreased in renal medulla of all groups (1.29 ± 0.28 vs 0.32 ± 0.11 and 0.94 ± 0.23 vs 0.20 ± 0.02, p < 0.01, respectively). Taken together, gonad dependent changes were seen in urine Na + and K + excretion and KKS behavior after ROMK blockade without changes in the BP and aldosterone. Moreover, the repression in Kcjn1 and Atp1α1 genes could be related to the observed ion transport changes.

In recent years, deep learning models have become the standard for agricultural computer vision. Such models are typically fine-tuned to agricultural tasks using model weights that were originally fit to more general, non-agricultural datasets. This lack of agriculture-specific fine-tuning potentially increases training time and resource use, and decreases model performance, leading to an overall decrease in data efficiency. To overcome this limitation, we collect a wide range of existing public datasets for 3 distinct tasks, standardize them, and construct standard training and evaluation pipelines, providing us with a set of benchmarks and pretrained models. We then conduct a number of experiments using methods that are commonly used in deep learning tasks but unexplored in their domain-specific applications for agriculture. Our experiments guide us in developing a number of approaches to improve data efficiency when training agricultural deep learning models, without large-scale modifications to existing pipelines. Our results demonstrate that even slight training modifications, such as using agricultural pretrained model weights, or adopting specific spatial augmentations into data processing pipelines, can considerably boost model performance and result in shorter convergence time, saving training resources. Furthermore, we find that even models trained on low-quality annotations can produce comparable levels of performance to their high-quality equivalents, suggesting that datasets with poor annotations can still be used for training, expanding the pool of currently available datasets. Our methods are broadly applicable throughout agricultural deep learning and present high potential for substantial data efficiency improvements.