Weed management technologies that can identify weeds and distinguish them from crops are in need of artificial intelligence solutions based on a computer vision approach, to enable the development precisely targeted autonomous robotic weed systems. A prerequisite such systems is create robust reliable object detection unambiguously food crops. One essential steps towards precision agriculture using annotated images train convolutional neural networks crops, which be later followed mechanical...

Differential privacy has recently gained prominence, especially in the context of private machine learning. While definition differential makes it possible to provably limit amount information leaked by an algorithm, practical implementations differentially algorithms often contain subtle vulnerabilities. Therefore, there is a need for effective methods that can audit before they are deployed real world. The article examines studies recommend guarantees It covers wide range topics on subject...

Segmentation neural networks are widely used in medical imaging to identify anomalies that may impact patient health. Despite their effectiveness, these face significant challenges, including the need for extensive annotated data, time-consuming manual segmentation processes and restricted data access due privacy concerns. In contrast, classification networks, similar capture essential parameters identifying objects during training. This paper leverages this characteristic, combined with...

This research delineates a pivotal advancement in the domain of sustainable energy systems, with focused emphasis on integration renewable sources—predominantly wind and solar power—into hydrogen production paradigm. At core this scientific endeavor is formulation implementation deep-learning-based framework for short-term localized weather forecasting, specifically designed to enhance efficiency derived from sources. The study presents comprehensive evaluation efficacy fully connected...

The integration of artificial intelligence (AI), particularly through machine learning (ML) and deep (DL) algorithms, marks a transformative progression in medical imaging diagnostics. This technical note elucidates novel methodology for semantic segmentation the vertebral column CT scans, exemplified by dataset 250 patients from Riga East Clinical University Hospital. Our approach centers on accurate identification labeling individual vertebrae, ranging C1 to sacrum–coccyx complex. Patient...

Object detection based on deep learning can be widely used in all kinds of agricultural applications. In this paper, we present a neural network (DNN) model for quince and raspberry RGB images. The trained DNN is YOLOv5 architecture it has 7 berry classes related to the development stage. provides sufficiently good performance precision trade-off. It useful process phenotyping agriculture experts, where yield size parameters have estimated. Using our shown that possible achieve mean Average...

With long-term changes in temperature and weather patterns, ecologically adaptable fruit varieties are becoming increasingly important agriculture. For selection of candidate cultivars breeding or for yield predictions, set characteristics at different growth stages need to be described evaluated, which is largely done visually. This a time-consuming labor-intensive process that also requires sufficient expert knowledge. The annotated dataset Japanese quince - QuinceSet consists images...

A specific signal digitising technique, based on sample value taking at time instants when the crosses a sinusoidal reference function, is considered. The various digital representations of original signals obtained in this way have invariable envelope and that leads to features unparalleled by other signals. timed sampling event sequences then actually could be considered as fully representing respective analog domain. Adding these collection more conventional types widen application range...

Features of digital signals, obtained in the case where signal sample values are taken at crossings an original analog and a sinusoidal reference function, discussed. It is shown that spectrum analysis this kind nonuniform with constant envelope not depending on parameters, might be performed without massive multiplication multi-digit numbers. Observed aliasing cross-interference effects considered adapting processing to specific signal-dependent sampling nonuniformities suggested, including...

Method for signal Analog-to-Event-to-Digital conversion using periodic sampling and precise event timing is described. It more suitable analogue information transmission of pulse width modulated signals with based demodulation than the referred to earlier considered methods this type it has also other advantages. This method described as an application widening functional capabilities high-performance Event Timer A033-ET system.

The RaspberrySet dataset is a valuable resource for those working in the field of agriculture, particularly selection and breeding ecologically adaptable berry cultivars. This because long-term changes temperature weather patterns have made it increasingly important crops to be able adapt their environment. To assess suitability different cultivars or make yield predictions, necessary describe evaluate berries’ characteristics at various growth stages. process typically carried out visually,...

In this technical note, we examine the capabilities of deep convolutional neural networks (DCNNs) for diagnosing osteoporosis through cone-beam computed tomography (CBCT) scans mandible. The evaluation was conducted using 188 patients’ mandibular CBCT images utilizing DCNN models built on ResNet-101 framework. We adopted a segmented three-phase method to assess osteoporosis. Stage 1 focused bone slice identification, 2 pinpointed coordinates cross-sectional views, and 3 bone’s thickness,...

The paper proposes an efficient method for training a neural network to count moving objects in video, while another concurrently prepares labeled dataset the first one. detection, tracking, and counting of is crucial effective Intelligence Transportation Systems (ITS), which should reduce congestion recognize traffic offenders on highways urban areas. Creation data one essential prerequisites successful application supervised machine learning. In this paper, experimental results automatic...

Model understanding is critical in many domains, particularly those involved high-stakes decisions, e.g., medicine, criminal justice, and autonomous driving. Explainable AI (XAI) methods are essential for working with black-box models such as convolutional neural networks. This paper evaluates the traffic sign classifier of Deep Neural Network (DNN) from Programmable Systems Intelligence Automobiles (PRYSTINE) project explainability. The results explanations were further used CNN PRYSTINE...

Hydrogen has the potential to revolutionize energy industry due its clean-burning and versatile properties. It is most abundant element in universe can be produced through a variety of methods, including electrolysis. The widespread adoption hydrogen faces various challenges, high cost production; thus, it important optimise production processes. This research focuses on development models for optimisation based external factors parameters. Models electricity prices are developed compared...

Approaches to representing in the frequency domain signals, reconstructed by using picosecond resolution Event Timers after compressive transmission of them, are investigated. The focus is on achieving high precision at complexity-reduced estimation Fourier coefficients performed rectangular basis functions instead sinusoidal. factors leading errors, degrading this type Discrete Transform and limiting range input signal, revealed. Developed two approaches significant suppressing errors discussed.

Bioimpedance signal analysis has to be performed in a rather wide frequency range. If fast ADCs, capable of taking sample values at very high sampling frequencies, are used for that, their bit rates typically low. It is suggested use pseudo-randomized quantizing and special methods handling the obtained data provide bioimpedance acquisition under required conditions demodulation multi-frequency carriers signals The described, computer simulated discussed.

Method for high precision analog input signal digitizing and compressive transmitting based on obtaining processing timing information by a picosecond resolution Event Timer is presented. While it has significant advantages, application of this type systems limited as the highest frequency signals cannot exceed half repetition rate involved periodic representative events formed timed Timer. An approach to eliminating limitation using deliberately randomized non-uniform sampling specific...

Standartised image compression/reconstruction algorithms are symmetric in the sense that computational complexities characterizing compression and reconstruction stages almost equal. An approach to assymetric is suggested discussed. Image performed according this extremely simple burden of commpression/reconstruction task shifted assymetrically stage. This approach, based on typical DASP methods, described The have been evaluated both basis computer simulations experimental studies obtained...