Search

Remote patient monitoring has emerged as a prominent non-invasive method, using digital technologies and computer vision (CV) to replace traditional invasive monitoring. While neonatal and pediatric departments embrace this approach, Pediatric Intensive Care Units (PICUs) face the challenge of occlusions hindering accurate image analysis and interpretation. Goal: In this study, we propose a hybrid approach to effectively segment common occlusions encountered in remote monitoring applications within PICUs. Our approach centers on creating a deep-learning pipeline for limited training data scenarios. Methods: First, a combination of the well-established Google DeepLabV3+ segmentation model with the transformer-based Segment Anything Model (SAM) is devised for occlusion segmentation mask proposal and refinement. We then train and validate this pipeline using a small dataset acquired from real-world PICU settings with a Microsoft Kinect camera, achieving an Intersection-over-Union (IoU) metric of 85%. Results: Both quantitative and qualitative analyses underscore the effectiveness of our proposed method. The proposed framework yields an overall classification performance with 92.5% accuracy, 93.8% recall, 90.3% precision, and 92.0% F1-score. Consequently, the proposed method consistently improves the predictions across all metrics, with an average of 2.75% gain in performance compared to the baseline CNN-based framework. Conclusions: Our proposed hybrid approach significantly enhances the segmentation of occlusions in remote patient monitoring within PICU settings. This advancement contributes to improving the quality of care for pediatric patients, addressing a critical need in clinical practice by ensuring more accurate and reliable remote monitoring.

Examining and potentially adjusting one's cognitive processes in response to dissatisfaction with one's performance is a fundamental aspect of intelligence. Remarkably, such sophisticated abstract concepts necessary for achieving Artificial General Intelligence can be effectively incorporated into basic Machine Learning algorithms. In this study, we introduce a method for replicating self-awareness through a supervisory Artificial Neural Network (ANN), which monitors patterns in the activation functions of an underlying ANN to identify signs of substantial uncertainty within the underlying ANN and, consequently, the reliability of its predictions. The underlying ANN in this context is a Convolutional Neural Network (CNN) ensemble primarily utilized for tasks related to facial recognition and facial expression analysis. We evaluate the performance of the supervisory ANNs using various activation functions as they learn to gauge the dependability of predictions made by the Inception v3 CNN ensemble. To conduct computational experiments, we employ a facial data set that incorporates makeup and occlusion factors. These experiments are designed to mimic real-world conditions where the training data set exclusively consists of images without makeup or occlusion, while the test data set comprises images featuring makeup and occlusion. This partitioning ensures the model is tested under challenging out-of-training data distribution scenarios. [ABSTRACT FROM AUTHOR]

Abstract Increasingly, automatic face recognition algorithms have become necessary with the development and extensive use of face recognition technology, particularly in the era of machine learning and artificial intelligence. However, the presence of unconstrained environmental conditions degrades the quality of acquired face images and may deteriorate the performance of many classical face recognition algorithms. Due to this backdrop, many researchers have given considerable attention to image restoration and enhancement mechanisms, but with minimal focus on occlusion-related and multiple-constrained problems. Although occlusion robust face recognition modules, via sparse representation have been explored, they require a large number of features to achieve correct computations and to maximize robustness to occlusions. Therefore, such an approach may become deficient in the presence of random occlusions of relatively moderate magnitude. This study assesses the robustness of Principal Component Analysis and Singular Value Decomposition using Discrete Wavelet Transformation for preprocessing and city block distance for classification (DWT-PCA/SVD-L1) face recognition module to image degradations due to random occlusions of varying magnitudes (10% and 20%) in test images acquired with varying expressions. Numerical evaluation of the performance of the DWT-PCA/SVD-L1 face recognition module showed that the use of the de-occluded faces for recognition enhanced significantly the performance of the study recognition module at each level (10% and 20%) of occlusion. The algorithm attained the highest recognition rate of 85.94% and 78.65% at 10% and 20% occlusions respectively, when the MICE de-occluded face images were used for recognition. With the exception of Entropy where MICE de-occluded face images attained the highest average value, the MICE and RegEM result in images of similar quality as measured by their Absolute mean brightness error (AMBE) and peak signal to noise ratio (PSNR). The study therefore recommends MICE as a suitable imputation mechanism for de-occlusion of face images acquired under varying expressions.

Tracking trajectories of the unconnected vehicles contributes to the improvement of traffic efficiency and safety. However, the effects of occlusions on the accuracy and reliability of the tracking results are nonnegligible. To address this issue, a modified multiple objects tracking algorithm was proposed to reduce the loss of trajectories caused by occlusions. The proposed algorithm was based on the multiple objects detection results, in which the motion states of the detected objects were determined. Further, the Kalman filter was employed to predict the trajectories, and each trajectory was uniquely matched to the detected object labeled with a tracking identity (ID) using the Hungarian algorithm. Afterward, a flexible life cycle strategy was proposed in terms of the speeds and accelerations of the detected objects, which controls the life cycle of the labeled trajectories when the occlusion occurred and guarantees the continuity of trajectories. The proposed tracking algorithm was tested on the Karlsruhe Institute of Technology and Toyota Technological Institute (KITTI) data set, and the classification of events, activities, and relationships workshops, multiple object tracking (CLEAR-MOT) metrics was introduced to evaluate the performance of the proposed algorithm. The results indicated that the proposed algorithm with appropriate life cycle apparently increased the precision and accuracy and reduced the influence of occlusions on multiple objects tracking. Future work will concentrate on field implementations of the algorithm, and various scenarios and weather conditions will be taken into account. [ABSTRACT FROM AUTHOR]

Using 10 years of satellite‐borne radar and lidar observations coupled with a novel method for automated occlusion identification, composite transects of cloud and precipitation across occluded thermal ridges of extratropical cyclones are, for the first time, constructed. These composites confirm that occluded sectors are characterized by the most extensive cloud cover and heaviest precipitation in any of the frontal regions of the cyclone. Hydrometeor frequency in occluded sectors is sensitive to the cyclone's ascent strength but not to the mean precipitable water in the cyclone's environment. This result is in contrast to the strong relationships between hydrometeor frequency and both precipitable water and ascent strength as previously reported in warm frontal regions. In both hemispheres, cloud and precipitation increase with the maximum value of the equivalent potential temperature at 700 hPa within the occluded thermal ridge, until a threshold is reached. For very large values of maximum equivalent potential temperature, hydrometeors become less frequent while precipitation rates increase. It is suggested that this conjunction is a by‐product of an increase in the frequency of convection in those instances. While in the Northern Hemisphere occluded sectors exhibit deeper and wider cloud structures than their Southern Hemisphere counterparts, their hydrometeor occurrence frequencies are less. The differences in maximum equivalent potential temperature of the thermal ridges in both hemispheres does not appear to explain the more frequent hydrometeors in the Southern Hemisphere. These relationships offer new perspectives on the interplay between cloud processes and cyclone evolution, as well as new observational constraints for process evaluation of Earth system models. [ABSTRACT FROM AUTHOR]

Accurate checkerboard detection is of vital importance for computer vision applications, and a variety of checkerboard detectors have been developed in the past decades. While some detectors are able to handle partially occluded checkerboards, they fail when a large occlusion completely divides the checkerboard. We propose a new checkerboard detection pipeline for occluded checkerboards that has a robust performance under varying levels of noise, blurring, and distortion, and for a variety of imaging modalities. This pipeline consists of a checkerboard detector and checkerboard enhancement with Gaussian processes (GP). By learning a mapping from local board coordinates to image pixel coordinates via a Gaussian process, we can fill in occluded corners, expand the board beyond the image borders, allocate detected corners that do not fit an initial grid, and remove noise on the detected corner locations. We show that our method can improve the performance of other publicly available state-of-the-art checkerboard detectors, both in terms of accuracy and the number of corners detected. Our code and datasets are made publicly available. The checkerboard detector pipeline is contained within our Python checkerboard detection library, called PyCBD. The pipeline itself is modular and easy to adapt to different use cases. [ABSTRACT FROM AUTHOR]

Recognition of degraded frontal face images acquired under occlusion constraints remain challenging despite the plethora of reconstruction mechanisms. Though recent works have leveraged on some imputation mechanisms in this regard, their robustness in multiple constrained environments may not be guaranteed and may be affected by the choice of pre-processing mechanism. This paper proposes enhancement mechanisms that augment or complement the use of three (3) multiple imputation mechanisms for facial reconstruction in the presence of multiple constraints (10% and 20% occlusions and varying facial expressions). Specifically, we propose the use of a Discrete Cosine Transform-based (DCT) denoising or a Discrete Wavelet-based denoising following Histogram Equalization (HE-DWT) of the reconstructed face images prior to recognition. Experimental results showed that the proposed augmented enhancements improved significantly the recognition rates (90.63% & 91.15% and 86.98% & 85.94% for DCT and HE-DWT at 10% and 20% occlusion levels respectively for Missforest de-occluded face images) as compared with DWT in recognizing degraded frontal face images under moderately low levels of occlusions and varying expressions.

Ego lane detection is one of the key techniques in Ego Lane Analysis System (ELAS) implemented in smart autonomous driving cars for lane detection in roads. This technique has been extensively studied in recent years because of its accurate and robust detection of shape and location of lanes. The conventional methods are less robust and computationally expensive since they have several challenges in localization of lanes due to presence of occlusions on roads. So to avoid these issues, this paper uses a novel 2-stage lane detection method using deep convolutional neural network to detect the lanes and its key-points by optimally fit a curve to the lane to compensate on above mentioned issues. The proposed methodology for lane detection predicts the key-points accurately and it robust under various weather conditions and highway driving scenarios. In terms of performance, this technique is fast and robust with low computational cost and has high performance when deployed on autonomous vehicle-based systems.

Accurate checkerboard detection is of vital importance for computer vision applications, and a variety of checkerboard detectors have been developed in the past decades. While some detectors are able to handle partially occluded checkerboards, they fail when a large occlusion completely divides the checkerboard. We propose a new checkerboard detection pipeline for occluded checkerboards that has a robust performance under varying levels of noise, blurring, and distortion, and for a variety of imaging modalities. This pipeline consists of a checkerboard detector and checkerboard enhancement with Gaussian processes (GP). By learning a mapping from local board coordinates to image pixel coordinates via a Gaussian process, we can fill in occluded corners, expand the board beyond the image borders, allocate detected corners that do not fit an initial grid, and remove noise on the detected corner locations. We show that our method can improve the performance of other publicly available state-of-the-art checkerboard detectors, both in terms of accuracy and the number of corners detected. Our code and datasets are made publicly available. The checkerboard detector pipeline is contained within our Python checkerboard detection library, called PyCBD. The pipeline itself is modular and easy to adapt to different use cases.

Vascular occlusions in patients with coronavirus diseases 2019 (COVID-19) have been frequently reported in severe outcomes mainly due to a dysregulation of neutrophils mediating neutrophil extracellular trap (NET) formation. Lung specimens from patients with COVID-19 have previously shown a dynamic morphology, categorized into three types of pleomorphic occurrence based on histological findings in this study. These vascular occlusions in lung specimens were also detected using native endogenous fluorescence or NEF in a label-free method. The three types of vascular occlusions exhibit morphology of DNA rich neutrophil elastase (NE) poor (type I), NE rich DNA poor (type II), and DNA and NE rich (type III) cohort of eleven patients with six males and five females. Age and gender have been presented in this study as influencing variables linking the occurrence of several occlusions with pleomorphic contents within a patient specimen and amongst them. This study reports the categorization of pleomorphic occlusions in patients with COVID-19 and the detection of these occlusions in a label-free method utilizing NEF. [ABSTRACT FROM AUTHOR]

Manual harvesting of coconuts is a highly risky and skill-demanding operation, and the population of people involved in coconut tree climbing has been steadily decreasing. Hence, with the evolution of tree-climbing robots and robotic end-effectors, the development of autonomous coconut harvesters with the help of machine vision technologies is of great interest to farmers. However, coconuts are very hard and experience high occlusions on the tree. Hence, accurate detection of coconut clusters based on their occlusion condition is necessary to plan the motion of the robotic end-effector. This study proposes a deep learning-based object detection Faster Regional-Convolutional Neural Network (Faster R-CNN) model to detect coconut clusters as non-occluded and leaf-occluded bunches. To improve identification accuracy, an attention mechanism was introduced into the Faster R-CNN model. The image dataset was acquired from a commercial coconut plantation during daylight under natural lighting conditions using a handheld digital single-lens reflex camera. The proposed model was trained, validated, and tested on 900 manually acquired and augmented images of tree crowns under different illumination conditions, backgrounds, and coconut varieties. On the test dataset, the overall mean average precision (mAP) and weighted mean intersection over union (wmIoU) attained by the model were 0.886 and 0.827, respectively, with average precision for detecting non-occluded and leaf-occluded coconut clusters as 0.912 and 0.883, respectively. The encouraging results provide the base to develop a complete vision system to determine the harvesting strategy and locate the cutting position on the coconut cluster. [ABSTRACT FROM AUTHOR]

Remote patient monitoring has emerged as a prominent non-invasive method, using digital technologies and computer vision (CV) to replace traditional invasive monitoring. While neonatal and pediatric departments embrace this approach, Pediatric Intensive Care Units (PICUs) face the challenge of occlusions hindering accurate image analysis and interpretation. Goal: In this study, we propose a hybrid approach to effectively segment common occlusions encountered in remote monitoring applications within PICUs. Our approach centers on creating a deep-learning pipeline for limited training data scenarios. Methods: First, a combination of the well-established Google DeepLabV3+ segmentation model with the transformer-based Segment Anything Model (SAM) is devised for occlusion segmentation mask proposal and refinement. We then train and validate this pipeline using a small dataset acquired from real-world PICU settings with a Microsoft Kinect camera, achieving an Intersection-over-Union (IoU) metric of 85%. Results: Both quantitative and qualitative analyses underscore the effectiveness of our proposed method. The proposed framework yields an overall classification performance with 92.5% accuracy, 93.8% recall, 90.3% precision, and 92.0% F1-score. Consequently, the proposed method consistently improves the predictions across all metrics, with an average of 2.75% gain in performance compared to the baseline CNN-based framework. Conclusions: Our proposed hybrid approach significantly enhances the segmentation of occlusions in remote patient monitoring within PICU settings. This advancement contributes to improving the quality of care for pediatric patients, addressing a critical need in clinical practice by ensuring more accurate and reliable remote monitoring., (© 2024 The Authors.)

Copyright of Canadian Journal of Plant Science is the property of Canadian Science Publishing and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

The need for accurate yield estimates for viticulture is becoming more important due to increasing competition in the wine market worldwide. One of the most promising methods to estimate the harvest is berry counting, as it can be approached non-destructively, and its process can be automated. In this article, we present a method that addresses the challenge of occluded berries with leaves to obtain a more accurate estimate of the number of berries that will enable a better estimate of the harvest. We use generative adversarial networks, a deep learning-based approach that generates a highly probable scenario behind the leaves exploiting learned patterns from images with non-occluded berries. Our experiments show that the estimate of the number of berries after applying our method is closer to the manually counted reference. In contrast to applying a factor to the berry count, our approach better adapts to local conditions by directly involving the appearance of the visible berries. Furthermore, we show that our approach can identify which areas in the image should be changed by adding new berries without explicitly requiring information about hidden areas.

Manual harvesting of coconuts is a highly risky and skill-demanding operation, and the population of people involved in coconut tree climbing has been steadily decreasing. Hence, with the evolution of tree-climbing robots and robotic end-effectors, the development of autonomous coconut harvesters with the help of machine vision technologies is of great interest to farmers. However, coconuts are very hard and experience high occlusions on the tree. Hence, accurate detection of coconut clusters based on their occlusion condition is necessary to plan the motion of the robotic end-effector. This study proposes a deep learning-based object detection Faster Regional-Convolutional Neural Network (Faster R-CNN) model to detect coconut clusters as non-occluded and leaf-occluded bunches. To improve identification accuracy, an attention mechanism was introduced into the Faster R-CNN model. The image dataset was acquired from a commercial coconut plantation during daylight under natural lighting conditions using a handheld digital single-lens reflex camera. The proposed model was trained, validated, and tested on 900 manually acquired and augmented images of tree crowns under different illumination conditions, backgrounds, and coconut varieties. On the test dataset, the overall mean average precision (mAP) and weighted mean intersection over union (wmIoU) attained by the model were 0.886 and 0.827, respectively, with average precision for detecting non-occluded and leaf-occluded coconut clusters as 0.912 and 0.883, respectively. The encouraging results provide the base to develop a complete vision system to determine the harvesting strategy and locate the cutting position on the coconut cluster.

• The framework enables enhanced UAV data acquisition for fruit detection purposes in vineyards considering the biophysical environment. • The optimization algorithm provides paths up to 24% shorter compared to conventional mission planners. • The proposed framework is designed for both single and multiple UAVs flying synchronously over the field. • The algorithm can be applied to woody crops other than vineyards trained in vertical trellis since they share similar biophysical characteristics and crop structures. Typically, commercial orchards and vineyards consist of large fields that encounter similar development phases at once. Thus, it becomes necessary to efficiently fly over all fields to detect fruit and identify their status in a very limited timeframe. For this purpose, Unmanned Aerial Vehicles (UAVs) path planning plays a pivotal role in agriculture as it enables optimal coverage of agricultural fields, leading to enhanced data acquisition and improved precision agriculture practices, for instance, disease assessment and pesticide application. In addition, deep learning techniques offer precise image analysis. On the one hand, object detection has been applied to agricultural fields to carry out a wide range of operations, such as detecting apples and predicting yield in vineyards, with higher detection accuracy when the fruits are fully visible. On the other hand, when crops present leaf-occlusion, the algorithms face difficulties and are unable to adapt to the specific characteristics of the field. Therefore, this study seeks to address this issue by developing a novel framework to enhance UAV path planning for data collection in vineyards, considering the current biophysical environment. To this end, the proposed framework requires two flights: i) a first flight (survey) to acquire insights on the crop structure and environment, and ii) a second flight using the Ant Colony Optimization Max-Min Ant System (ACO-MMAS) algorithm to enhance image acquisition by considering multiple angles to overcome partial leaf-occlusion. Further, the optimisation algorithm can potentially boost the acquisition of datasets for fruit detection by considering single and multiple UAVs flying synchronously while ensuring a safe distance between platforms and efficient coverage. The method was tested in two vineyards with different environmental characteristics, increasing levels of difficulty and acquired during two different growing seasons. It improved the length of the computed paths by up to 24%, compared to a base algorithm that considers only the closest point without any optimisation, improving the decision-making processes and resource allocation in crop management. [ABSTRACT FROM AUTHOR]

Peripheral artery disease (PAD) is the buildup of calcium and fatty deposits in the arterial walls (atherosclerosis). This is an important clinical issue, specifically in cases with multilevel lesions. A patient underwent sequential angioplasty treatment for major PAD, which was characterized by multilevel lesions affecting both the infrapopliteal arteries. The proximal vessels and infrapopliteal vessels are mostly observed to be affected by PAD, thus the patient likely has PAD localized to the lower leg. In the femoropopliteal segment, lower extremity artery or aortic atherosclerotic occlusive disease can lead to significant outcomes. Severe claudication and pain during rest in both legs were observed in a patient with a history of hypertension and diabetes mellitus. With an angiography, the superficial femoral, popliteal, and tibial arteries have been shown to have major stenoses and occlusions. A progressive treatment was used because of the complexity of the lesions initiating with endovascular revascularization of the superficial femoral artery. The popliteal and tibial arteries were then repaired with angioplasty and stent placement. After the treatment, the patient's symptoms significantly improved, including elimination of their rest discomfort and claudication. Measurements of the ankle-brachial index (ABI) indicated that the affected limbs' perfusion was refined. Six months later, a follow-up angiography revealed intact vessels with no restenosis. This case report shows the successful outcome of recurrent angioplasty in curing complicated multilevel PAD, giving symptomatic relief and maintaining limb perfusion. This research is required to assess the long-term outcomes and longevity of this kind of treatment in patient populations that are comparable to others., Competing Interests: The authors have declared that no competing interests exist., (Copyright © 2024, Patil et al.)

• Santiago Urban Dataset SUD was acquired with MLS and HMLS equipment. • Occlusions are reduced by the integration of MLS and HMLS data. • Point clouds are labelled by heuristic and Deep Learning methods in eight classes. • SUD is valid for comparing semantic segmentation works. • SUD is valid for urban 3D mobility studies. Santiago Urban Dataset SUD is a real dataset that combines Mobile Laser Scanning (MLS) and Handheld Mobile Laser Scanning (HMLS) point clouds. The data is composed by 2 km of streets, sited in Santiago de Compostela (Spain). Point clouds undergo a manual labelling process supported by both heuristic and Deep Learning methods, resulting in the classification of eight specific classes: road, sidewalk, curb, buildings, vehicles, vegetation, poles , and others. Three PointNet++ models were trained; the first one using MLS point clouds, the second one with HMLS point clouds and the third one with both H&MLS point clouds. In order to ascertain the quality and efficacy of each Deep Learning model, various metrics were employed, including confusion matrices, precision, recall, F1-score, and IoU. The results are consistent with other state-of-the-art works and indicate that SUD is valid for comparing point cloud semantic segmentation works. Furthermore, the survey's extensive coverage and the limited occlusions indicate the potential utility of SUD in urban mobility research. [ABSTRACT FROM AUTHOR]

Nuclei segmentation plays a crucial role in disease understanding and diagnosis. In whole slide images, cell nuclei often appear overlapping and densely packed with ambiguous boundaries due to the underlying 3D structure of histopathology samples. Instance segmentation via deep neural networks with object clustering is able to detect individual segments in crowded nuclei but suffers from a limited field of view, and does not support amodal segmentation. In this work, we introduce a dense feature pyramid network with a feature mixing module to increase the field of view of the segmentation model while keeping pixel-level details. We also improve the model output quality by adding a multi-scale self-attention guided refinement module that sequentially adjusts predictions as resolution increases. Finally, we enable clusters to share pixels by separating the instance clustering objective function from other pixel-related tasks, and introduce supervision to occluded areas to guide the learning process. For evaluation of amodal nuclear segmentation, we also update prior metrics used in common modal segmentation to allow the evaluation of overlapping masks and mitigate over-penalization issues via a novel unique matching algorithm. Our experiments demonstrate consistent performance across multiple datasets with significantly improved segmentation quality., Competing Interests: Declaration of competing interest None Declared., (Copyright © 2024 Elsevier Ltd. All rights reserved.)

Through biomineralization, calcareous composites are produced with exceptional properties, evolution-optimized for specific function. The bioinspired quest to understand how properties are controlled and enhanced is motivated by their fundamental and technological significance. The incorporation of small molecules and/or biopolymers as inter- and intra-crystalline additives in the CaCO 3 matrix, is widely employed by organisms to achieve diverse functions. The interactions between the components during the early events within the precipitation medium, and when entrapped through precipitation-crystallization, are key players of process–property regulation. In addition to identifying the bulk matrices and the incorporated molecules, we show how solid-state NMR methods are tailored to directly report the chemical-structural details of the inorganic interface that surrounds an occlusion. Solid-state NMR is uniquely suited for that and is applicable to stable or spontaneously transforming lattices, crystalline or amorphous. Our findings are grouped to highlight the connection between the molecular level and tunability of macroscopic properties. [Display omitted] [ABSTRACT FROM AUTHOR]

In multi-target tracking, object interactions and occlusions are two significant factors that affect tracking performance. To settle this, we propose an identity association network (IANet) that integrates the geometry refinement network (GRNet) and the identity verification (IV) module to perform data association and reason the mapping between the detections and tracklets. In our data association process, the object drifts caused by object interactions are suppressed effectively by encoding the direction and velocity of objects to refine the geometric position of tracklets. The tracklets with refined geometric information are further utilized in the IV module to achieve a sufficient encoding of multivariate spatial cues including both appearance and geometry information, which defeats the misleading impacts of interactions and occlusions dramatically in multi-object tracking. The extensive experiments and comparative evaluations have demonstrated that our proposed method can significantly outperform many state-of-the-art methods on benchmarks of 2D MOT2015, MOT16, MOT17, MOT20, and KITTI by using public detection and online settings. [ABSTRACT FROM AUTHOR]

The need for accurate yield estimates for viticulture is becoming more important due to increasing competition in the wine market worldwide. One of the most promising methods to estimate the harvest is berry counting, as it can be approached non-destructively, and its process can be automated. In this article, we present a method that addresses the challenge of occluded berries with leaves to obtain a more accurate estimate of the number of berries that will enable a better estimate of the harvest. We use generative adversarial networks, a deep learning-based approach that generates a highly probable scenario behind the leaves exploiting learned patterns from images with non-occluded berries. Our experiments show that the estimate of the number of berries after applying our method is closer to the manually counted reference. In contrast to applying a factor to the berry count, our approach better adapts to local conditions by directly involving the appearance of the visible berries. Furthermore, we show that our approach can identify which areas in the image should be changed by adding new berries without explicitly requiring information about hidden areas., Competing Interests: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest., (Copyright © 2022 Kierdorf, Weber, Kicherer, Zabawa, Drees and Roscher.)

Wearing protective masks has become commonplace in several countries as a result of the epidemic connected to the Covid-19 virus. Wearing a mask obscures a considerable part of the face, making some facial recognition techniques difficult to complete and obstructing the operation of various identifying systems, such as access control systems. In this paper, we offer an original approach that allows many face recognition systems to continue to identify persons even when wearing protective masks. The proposed approach is mainly based on the prior use of skin detection techniques. We validated our method using the Eigenfaces method by the FEI database, which we supplemented with faces wearing protective masks. The evaluation results of our approach are very satisfactory., (© 2019, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd.)