Brno urban dataset: Winter extension.

This paper presents our latest extension of the Brno Urban Dataset (BUD), the Winter Extension (WE). The dataset contains data from commonly used sensors in the automotive industry, like four RGB and single IR cameras, three 3D LiDARs, differential RTK GNSS receiver with heading estimation, the IMU and FMCW radar. Data from all sensors are precisely timestamped for future offline interpretation and data fusion. The most significant gain of the dataset is the focus on the winter conditions in snow-covered environments. Only a few public datasets deal with these kinds of conditions. We recorded the dataset during February 2021 in Brno, Czechia, when fresh snow covers the entire city and the surrounding countryside. The dataset contains situations from the city center, suburbs, highways as well as the countryside. Overall, the new extension adds three hours of real-life traffic situations from the mid-size city to the existing 10 h of original records. Additionally, we provide the precalculated YOLO neural network object detection annotations for all five cameras for the entire old data and the new ones. The dataset is suitable for developing mapping and navigation algorithms as well as the collision and object detection pipelines. The entire dataset is available as open-source under the MIT license.

Visual Diagnosis of the Varroa Destructor Parasitic Mite in Honeybees Using Object Detector Techniques.

The Varroa destructor mite is one of the most dangerous Honey Bee (Apis mellifera) parasites worldwide and the bee colonies have to be regularly monitored in order to control its spread. In this paper we present an object detector based method for health state monitoring of bee colonies. This method has the potential for online measurement and processing. In our experiment, we compare the YOLO and SSD object detectors along with the Deep SVDD anomaly detector. Based on the custom dataset with 600 ground-truth images of healthy and infected bees in various scenes, the detectors reached the highest F1 score up to 0.874 in the infected bee detection and up to 0.714 in the detection of the Varroa destructor mite itself. The results demonstrate the potential of this approach, which will be later used in the real-time computer vision based honey bee inspection system. To the best of our knowledge, this study is the first one using object detectors for the Varroa destructor mite detection on a honey bee. We expect that performance of those object detectors will enable us to inspect the health status of the honey bee colonies in real time.

Fully Automated DCNN-Based Thermal Images Annotation Using Neural Network Pretrained on RGB Data.

One of the biggest challenges of training deep neural network is the need for massive data annotation. To train the neural network for object detection, millions of annotated training images are required. However, currently, there are no large-scale thermal image datasets that could be used to train the state of the art neural networks, while voluminous RGB image datasets are available. This paper presents a method that allows to create hundreds of thousands of annotated thermal images using the RGB pre-trained object detector. A dataset created in this way can be used to train object detectors with improved performance. The main gain of this work is the novel method for fully automatic thermal image labeling. The proposed system uses the RGB camera, thermal camera, 3D LiDAR, and the pre-trained neural network that detects objects in the RGB domain. Using this setup, it is possible to run the fully automated process that annotates the thermal images and creates the automatically annotated thermal training dataset. As the result, we created a dataset containing hundreds of thousands of annotated objects. This approach allows to train deep learning models with similar performance as the common human-annotation-based methods do. This paper also proposes several improvements to fine-tune the results with minimal human intervention. Finally, the evaluation of the proposed solution shows that the method gives significantly better results than training the neural network with standard small-scale hand-annotated thermal image datasets.

The RoScan Thermal 3D Body Scanning System: Medical Applicability and Benefits for Unobtrusive Sensing and Objective Diagnosis.

The RoScan is a novel, high-accuracy multispectral surface scanning system producing colored 3D models that include a thermal layer. (1) Background: at present, medicine still exhibits a lack of objective diagnostic methods. As many diseases involve thermal changes, thermography may appear to be a convenient technique for the given purpose; however, there are three limiting problems: exact localization, resolution vs. range, and impossibility of quantification. (2) Methods: the basic principles and benefits of the system are described. The procedures rely on a robotic manipulator with multiple sensors to create a multispectral 3D model. Importantly, the structure is robust, scene-independent, and features quantifiable measurement uncertainty; thus, all of the above problems of medical thermography are resolved. (3) Results: the benefits were demonstrated by several pilot case studies: medicament efficacy assessment in dermatology, objective recovery progress assessment in traumatology, applied force quantification in forensic sciences, exact localization of the cause of pain in physiotherapy, objective assessment of atopic dermatitis, and soft tissue volumetric measurements. (4) Conclusion: the RoScan addresses medical quantification, which embodies a frequent problem in several medical sectors, and can deliver new, objective information to improve the quality of healthcare and to eliminate false diagnoses.

LOCALIZATION OF IONIZING RADIATION SOURCES VIA AN AUTONOMOUS ROBOTIC SYSTEM.

The article discusses an autonomous and flexible robotic system for radiation monitoring. The detection part of the system comprises two NaI(Tl) scintillation detectors: one of these is collimated to allow directionally sensitive measurements and the other is used to calculate the dose rate and provides sufficient sensitivity. Special algorithms for autonomous operation of an unmanned ground vehicle were developed, utilizing radiation characteristics acquired by the implemented detection system. The system was designed to operate in three modes: radiation mapping, localization of discrete sources and inspection of a region of interest. All of the modes were verified experimentally. In the localization mode, the time required to localize ionizing radiation sources was reduced by half compared to the field mapping mode exploiting parallel trajectories; the localization accuracy remained the same. In the inspection mode, the desired functionality was achieved, and the changes in the sources arrangement were detected reliably in the experiments.

Towards Automatic UAS-Based Snow-Field Monitoring for Microclimate Research.

This article presents unmanned aerial system (UAS)-based photogrammetry as an efficient method for the estimation of snow-field parameters, including snow depth, volume, and snow-covered area. Unlike similar studies employing UASs, this method benefits from the rapid development of compact, high-accuracy global navigation satellite system (GNSS) receivers. Our custom-built, multi-sensor system for UAS photogrammetry facilitates attaining centimeter- to decimeter-level object accuracy without deploying ground control points; this technique is generally known as direct georeferencing. The method was demonstrated at Mapa Republiky, a snow field located in the Krkonose, a mountain range in the Czech Republic. The location has attracted the interest of scientists due to its specific characteristics; multiple approaches to snow-field parameter estimation have thus been employed in that area to date. According to the results achieved within this study, the proposed method can be considered the optimum solution since it not only attains superior density and spatial object accuracy (approximately one decimeter) but also significantly reduces the data collection time and, above all, eliminates field work to markedly reduce the health risks associated with avalanches.

Breath Analysis Using a Time-of-Flight Camera and Pressure Belts.

The proper way of breathing is important for everyone. Healthy people often do not follow respiration until breathing problems start-during stress or during sport activity in physiological cases. More serious cases are stroke, injury, or surgery of the chest and others. So, learning to breathe correctly and/or breathing diagnosis is considerable for many reasons. Two novel methods of breath analysis suitable for diagnostics and rehabilitation are presented. The first technique utilizes pressure belts fastened to the patient's belly and chest, and the second method relies on a SwissRanger SR-4000 time-of-flight camera. The measurement principles are described together with the advantages and disadvantages of the applied techniques. The SwissRanger camera depth calibration is proposed to facilitate better results during the breath analysis. The methods are tested on a group of students to provide a comparison of their individual performances. As it was demonstrated, presented methods proved to work reliably. The method based on time-of-flight camera seems to be more suitable for diagnosis, while the method based on pressure belts is more suitable for rehabilitation and biofeedback applications.

Limb volume measurements: comparison of accuracy and decisive parameters of the most used present methods.

Limb volume measurements are used for evaluating growth of muscle mass and effectivity of strength training. Beside sport sciences, it is used e.g. for detection of oedemas, lymphedemas or carcinomas or for examinations of muscle atrophy. There are several commonly used methods, but there is a lack of clear comparison, which shows their advantages and limits. The accuracy of each method is uncertainly estimated only. The aim of this paper is to determine and experimentally verify their accuracy and compare them among each other. Water Displacement Method (WD), three methods based on circumferential measures-Frustum Sign Model (FSM), Disc Model (DM), Partial Frustum Model (PFM) and two 3D scan based methods Computed Tomography (CT) and Magnetic Resonance Imaging (MRI) were compared. Precise reference cylinders and limbs of two human subjects were measured 10 times by each method. Personal dependency of methods was also tested by measuring 10 times the same object by 3 different people. Accuracies: WD 0.3 %, FSM 2-8 % according person, DM, PFM 1-8 %, MRI 2 % (hand) or 8 % (finger), CT 0.5 % (hand) or 2 % (finger);times: FSM 1 min, CT 7 min, WD, DM, PFM 15 min, MRI 19 min; and more. WD was found as the best method for most of uses with best accuracy. The CT disposes with almost the same accuracy and allows measurements of specific regions (e.g. particular muscles), as same as MRI, which accuracy is worse though, but it is not harmful. Frustum Sign Model is usable for very fast estimation of limb volume, but with lower accuracy, Disc Model and Partial Frustum Model is useful in cases when Water Displacement cannot be used.

Remote-controlled robotic platform ORPHEUS as a new tool for detection of bacteria in the environment.

Remote-controlled robotic systems are being used for analysis of various types of analytes in hostile environment including those called extraterrestrial. The aim of our study was to develop a remote-controlled robotic platform (ORPHEUS-HOPE) for bacterial detection. For the platform ORPHEUS-HOPE a 3D printed flow chip was designed and created with a culture chamber with volume 600 µL. The flow rate was optimized to 500 µL/min. The chip was tested primarily for detection of 1-naphthol by differential pulse voltammetry with detection limit (S/N = 3) as 20 nM. Further, the way how to capture bacteria was optimized. To capture bacterial cells (Staphylococcus aureus), maghemite nanoparticles (1 mg/mL) were prepared and modified with collagen, glucose, graphene, gold, hyaluronic acid, and graphene with gold or graphene with glucose (20 mg/mL). The most up to 50% of the bacteria were captured by graphene nanoparticles modified with glucose. The detection limit of the whole assay, which included capturing of bacteria and their detection under remote control operation, was estimated as 30 bacteria per µL.

Robotic 3D scanner as an alternative to standard modalities of medical imaging.

There are special medical cases, where standard medical imaging modalities are able to offer sufficient results, but not in the optimal way. It means, that desired results are produced with unnecessarily high expenses, with redundant informations or with needless demands on patient. This paper deals with one special case, where information useful for examination is the body surface only, inner sight into the body is needless. New specialized medical imaging device is developed for this situation. In the Introduction section, analysis of presently used medical imaging modalities is presented, which declares, that no available imaging device is best fitting for mentioned purposes. In the next section, development of the new specialized medical imaging device is presented, and its principles and functions are described. Then, the parameters of new device are compared with present ones. It brings significant advantages comparing to present imaging systems.