Target Box Research Articles

Research on Flower Image Classification Method Based on YOLOv5

The rapid development of deep learning has accelerated the progress of related technologies in the computer vision field and it has broad application prospects. Due to flower inter-class similarity and intra-class differences, flower image classification has essential research value. To achieve flower image classification, this paper proposes a deep learning method using the current powerful object detection algorithm YOLOv5 to achieve fine-grained image classification of flowers. Overlap and occluded objects often appear in the images of the flowers, so the DIoU_NMS algorithm is used to select the target box to enhance the detection of the blocked objects. The experimental dataset comes from the Kaggle platform, and experimental results show that the proposed model in this paper can effectively identify five types of flowers contained in the dataset, Precision reaching 0.942, Recall reaching 0.933, and mAP reaching 0.959. Compared with YOLOv3 and Faster-RCNN, this model has high recognition accuracy, real-time performance, and good robustness. The mAP of this model is 0.051 higher than the mAP of YOLOv3 and 0.102 higher than the mAP of Raster-RCNN.

Optimized loss functions for object detection and application on nighttime vehicle detection

The loss function is a crucial factor that affects the detection precision in the object detection task. In this paper, we optimize both two loss functions for classification and localization simultaneously. Firstly, we reconstruct the classification loss function by combining the prediction results of localization, aiming to establish the correlation between localization and classification subnetworks. Compared to the existing studies, in which the correlation is only established among the positive samples and applied to improve the localization accuracy of predicted boxes, this paper utilizes the correlation to define the hard negative samples and then puts emphasis on the classification of them. Thus the whole misclassified rate for negative samples can be reduced. Besides, a novel localization loss named MIoU is proposed by incorporating a Mahalanobis distance between the predicted box and target box, eliminating the gradients inconsistency problem in the DIoU loss, further improving the localization accuracy. Finally, the proposed methods are applied to train the networks for nighttime vehicle detection. Experimental results show that the detection accuracy can be outstandingly improved with our proposed loss functions without hurting the detection speed.

Metformin prevents p-tau and amyloid plaque deposition and memory impairment in diabetic mice.

Insulin deficiency or resistance can promote dementia and hallmarks of Alzheimer's disease (AD). The formation of neurofibrillary tangles of p-TAU protein, extracellular Aβ plaques, and neuronal loss is related to the switching off insulin signaling in cognition brain areas. Metformin is a biguanide antihyperglycemic drug used worldwide for the treatment of type 2 diabetes. Some studies have demonstrated that metformin exerts neuroprotective, anti-inflammatory, anti-oxidant, and nootropic effects. This study aimed to evaluate metformin's effects on long-term memory and p-Tau and amyloid β modulation, which are hallmarks of AD in diabetic mice. Swiss Webster mice were distributed in the following experimental groups: control; treated with streptozotocin (STZ) that is an agent toxic to the insulin-producing beta cells; STZ + metformin 200mg/kg (M200). STZ mice showed significant augmentation of time spent to reach the target box in the Barnes maze, while M200 mice showed a significant time reduction. Moreover, the M200 group showed reduced GFAP immunoreactivity in hippocampal dentate gyrus and CA1 compared with the STZ group. STZ mice showed high p-Tau levels, reduced p-CREB, and accumulation of β-amyloid (Aβ) plaque in hippocampal areas and corpus callosum. In contrast, all these changes were reversed in the M200 group. Protein expressions of p-Tau, p-ERK, pGSK3, iNOS, nNOS, PARP, Cytochrome c, caspase 3, and GluN2A were increased in the parietal cortex of STZ mice and significantly counteracted in M200 mice. Moreover, M200 mice also showed significantly high levels of eNOS, AMPK, and p-AKT expression. In conclusion, metformin improved spatial memory in diabetic mice, which can be associated with reducing p-Tau and β-amyloid (Aβ) plaque load and inhibition of neuronal death.

Object detection/tracking toward underwater photographs by remotely operated vehicles (ROVs)

Owing to the shortcoming of image blur, scale diversity, and complex backgrounds, underwater target detection and tracking is a tough challenge in real-world ROVs applications. In this paper, an underwater target detection method based on the class-weighted YOLO network is proposed. The key technique is to construct the class-weighted loss function based on the deep network YOLO to balance the sample difficulty and achieve better results. In addition, an adaptive dimensional clustering algorithm of target box is introduced to further improve the detection performance. Firstly, according to the imaging characteristics of the small underwater target in the side-scan sonar image, the sonar image is filtered and segmented by clustering to significantly reduce noises in each image. Afterward, the speckle detection is deployed to extract the suspected target area in the side-scan sonar image. Finally, the sonar image is segmented according to the automatic threshold segmentation and receive the binary image of the target region. The target scale is estimated by the second moment and the false target is eliminated. Correspondingly, the accurate detection of small underwater target can be achieved. The moving object pixel is detected by background difference and three-frame difference respectively. On this basis, we combine the background difference and the three-frame difference. The background difference characterizes the objects capturing the incomplete message detected by the three-frame difference. Finally, morphological operation is leveraged to eliminate the noise caused by non-static background objects. Experiments on multiple ROVs have validated that, the proposed algorithm can improve the average accuracy and the recall rate remarkably in the dense target detection task that contains 20 targets averagely.

Cyclotron production of 225Ac from an electroplated 226Ra target

PurposeWe demonstrate cyclotron production of high-quality 225Ac using an electroplated 226Ra target.Methods226Ra was extracted from legacy Ra sources using a chelating resin. Subsequent ion-exchange purification gave pure 226Ra with a certain amount of carrier Ba. The radium target was prepared by electroplating. We successfully deposited about 37 MBq of 226Ra on a target box. Maximum activation was achieved using 15.6 MeV protons on the target at 20 µA for 5 h. Two functional resins with various concentrations of nitric acid purified 225Ac and recovered 226Ra. Cooling the intermediate 225Ac for 2–3 weeks decayed the major byproduct of 226Ac and increased the radionuclidic purity of 225Ac. Repeating the same separation protocol provided high-quality 225Ac.ResultsWe obtained 225Ac at a yield of about 2.4 MBq at the end of bombardment (EOB), and the subsequent initial purification gave 1.7 MBq of 225Ac with 226Ac/225Ac ratio of < 3% at 4 days from EOB. Additional cooling time coupled with the separation procedure (secondary purification) effectively increased the 225Ac (4n + 1 series) radionuclidic purity up to 99 + %. The recovered 225Ac had a similar identification to commercially available 225Ac originating from a 229Th/225Ac generator.ConclusionThis procedure, which involves the 226Ra(p,2n)225Ac reaction and the appropriate purification, has the potential to be a major alternative pathway for 225Ac production because it can be performed in any facility with a compact cyclotron to address the increasing demand for 225Ac.

Phosphatase and Tensin Homolog Deleted on Chromosome Ten Knockdown Attenuates Cognitive Deficits by Inhibiting Neuroinflammation in a Mouse Model of Perioperative Neurocognitive Disorder

Phosphatase and tensin homolog deleted on chromosome ten (PTEN) is a crucial regulator of neuronal development, neuronal survival, axonal regeneration, and synaptic plasticity. In this study we examined the potential role of PTEN in cognitive function in a mouse model of perioperative neurocognitive disorder (PND). Adult male C57BL/6J mice received intracerebroventricular injections of small interfering RNA (siRNA) against PTEN or control siRNA 3 days prior to exploratory laparotomy (n = 8 per group). A group of healthy mice not undergoing surgery included as additional control. Barnes maze and fear conditioning tests were conducted 7 days after surgery. Mice were then sacrificed to examine the expression of PTEN, AMP-activated protein kinase (AMPK), ionized calcium binding adaptor molecule (Iba)-1, B-cell lymphoma (Bcl)-2, Bcl2-associated X protein (Bax), interleukin (IL)-1β, and tumor necrosis factor (TNF)-α in the hippocampus. The microglial activation was examined by immunohistochemistry using Iba-1 as a microglia maker. Nissl and terminal transferase deoxyuridine triphosphate nick-end labeling (TUNEL) staining were used to measure cell death and apoptosis. In comparison to the healthy controls, surgically treated mice had longer latency to identify the target box in both training and testing sessions in the Barnes maze test and shorter freezing time in the fear conditioning test. Surgically treated mice had increased expression of PTEN, AMPK, Bax, IL-1β, and TNF-α, as well as increasing number of activated microglia and apoptosis neurons in the hippocampus. PTEN knockdown significantly attenuated the behavioral deficits in Barnes maze and fear conditioning tests, as well as over-expression of PTEN, AMPK, Bax, IL-1β, and TNF-α induced by surgery. PTEN knockdown could attenuate cognitive deficits induced by trauma, likely through inhibiting the activation of microglia.

A multivariate intersection over union of SiamRPN network for visual tracking

SiamPRN algorithm performs well in visual tracking, but it is easy to drift under occlusion and fast motion scenes because it uses ell _1-smooth loss function to measure the regression location of bounding box. In this paper, we propose a multivariate intersection over union (MIOU) loss in SiamRPN tracking framework. Firstly, MIOU loss includes three geometric factors in regression: the overlap area ratio, the center distance ratio, and the aspect ratio, which can better reflect the coincidence degree of target box and prediction box. Secondly, we improve the definition of aspect ratio loss to avoid gradient explosion, improve the optimization performance of prediction box. Finally, based on SiamPRN tracker, we compared the tracking performance of ell _1-smooth loss, IOU loss, GIOU loss, DIOU loss, and MIOU loss. Experimental results show that the MIOU loss has better target location regression than other loss functions on the OTB2015 and VOT2016 benchmark, especially for the challenges of occlusion, illumination change and fast motion.

2020 NICE guideline update: Good news for pregnant women with type 1 diabetes and past or current gestational diabetes.

In December 2020, the National Institute for Health and Care Excellence (NICE) reviewed the evidence and updated their recommendations on intermittently scanned (commonly known as Flash) and Continuous Glucose Monitoring (CGM) during pregnancy for women with type 1 diabetes (1). The NICE guidelines now recommend offering CGM to all pregnant women with type 1 diabetes to help them meet their pregnancy glucose targets and improve neonatal outcomes. Their evidence review, based on the CONCEPTT randomised trial (2) and a Swedish observational study (3) found that, compared to capillary glucose monitoring, CGM resulted in more women achieving their blood glucose targets, fewer caesarean sections and fewer neonatal intensive care admissions.

Role of Sox2 in Learning, Memory, and Postoperative Cognitive Dysfunction in Mice.

Postoperative cognitive dysfunction (POCD) is a significant clinical issue. Its neuropathogenesis has not been clearly identified and effective interventions for clinical use to reduce POCD have not been established. This study was designed to determine whether environmental enrichment (EE) or cognitive enrichment (CE) reduces POCD and whether sex-determining region Y-box-2 regulated by sirtuin 1, plays a role in the effect. Eighteen-month-old male mice were subjected to right-common-carotid-artery exposure under sevoflurane anesthesia. Some of them stayed in cages with EE or CE after the surgery. Learning and memory of mice were tested by a Barnes maze and fear conditioning, starting 2 weeks after the surgery. Sex-determining region Y-box-2 (Sox2) in the brain was silenced by small hairpin RNA (shRNA). Immunofluorescent staining was used to quantify Sox2-positive cells. Surgery reduced Sox2-positive cells in the hippocampus (64 ± 9 cells vs. 91 ± 9 cells in control group, n = 6, p < 0.001) and impaired learning and memory (time to identify target box one day after training sessions in the Barnes maze test: 132 ± 53 s vs. 79 ± 53 s in control group, n = 10, p = 0.040). EE or CE applied after surgery attenuated this reduction of Sox2 cells and POCD. Surgery reduced sirtuin 1 activity and CE attenuated this reduction. Resveratrol, a sirtuin 1 activator, attenuated POCD and surgery-induced decrease of Sox2-positive cells. Silencing shRNA reduced the Sox2-positive cells in the hippocampus and impaired learning and memory in mice without surgery. These results suggest a role of Sox2 in learning, memory, and POCD. EE and CE attenuated POCD via maintaining Sox2-positive cells in the hippocampus.

A New Semantic-Based Tool Detection Method for Robots

Home helper robots have become more acceptable due to their excellent image recognition ability. However, some common household tools remain challenging to recognize, classify, and use by robots. We designed a detection method for the functional components of common household tools based on the mask regional convolutional neural network (Mask-R-CNN). This method is a multitask branching target detection algorithm that includes tool classification, target box regression, and semantic segmentation. It provides accurate recognition of the functional components of tools. The method is compared with existing algorithms on the dataset UMD Part Affordance dataset and exhibits effective instance segmentation and key point detection, with higher accuracy and robustness than two traditional algorithms. The proposed method helps the robot understand and use household tools better than traditional object detection algorithms.

An Automatically Initialized Level Set Method for Ochotona Curzoniae Image Segmentation

In order to effectively solve this problem that that the traditional level set method needs to mark the initial contour manually, a novel method that automatic initialization by utilizing deep learning is proposed in this paper. Firstly, a target detection network based on Faster-RCNN is used for target detection of ochotona curzoniae. Secondly, the detected target boxes are regarded as the initial contour of level set models, and ochotona curzoniae images are segmented by level set models, by which the problem of manually plotting the initial contour required by traditional level set models is solved. The experimental results of ochotona curzoniae image segmentation show that there is little difference between the level set segmentation method in which the initial contour of the target is automatically generated by deep learning(LSECA) is used to automatically target initial contour and the level set segmentation method where initial contour needs to be manually plotted. Therefore, the experimental results prove that it is feasible to automatically obtain the target initial contour through deep learning.

Algorithm of Helmet Wearing Detection Based on AT-YOLO Deep Mode

The existing safety helmet detection methods are mainly based on one-stage object detection algorithms with high detection speed to reach the real-time detection requirements, but they can’t accurately detect small objects and objects with obstructions. Therefore, we propose a helmet detection algorithm based on the attention mechanism (AT-YOLO). First of all, a channel attention module is added to the YOLOv3 backbone network, which can adaptively calibrate the channel features of the direction to improve the feature utilization, and a spatial attention module is added to the neck of the YOLOv3 network to capture the correlation between any positions in the feature map so that to increase the receptive field of the network. Secondly, we use DIoU (Distance Intersection over Union) bounding box regression loss function, it not only improving the measurement of bounding box regression loss but also increases the normalized distance loss between the prediction boxes and the target boxes, which makes the network more accurate in detecting small objects and faster in convergence. Finally, we explore the training strategy of the network model, which improves network performance without increasing the inference cost. Experiments show that the mAP of the proposed method reaches 96.5%, and the detection speed can reach 27 fps. Compared with other existing methods, it has better performance in detection accuracy and speed.

Real-time assessment of looking time at central environmental cues for spontaneous recall in 35-month-olds

Evidence supporting the assumption that spontaneous recall of previously experienced events is triggered by distinct, environmental cues has hitherto exclusively been based on post-hoc reasoning and not on direct, in vivo examination. Using a novel experimental paradigm, we present the first real-time analyses of 35-month-old’s looking time as spontaneous retrieval unfolds. Twenty-nine middle-to-higher SES children (out of 114 participating) spontaneously recalled having taken part in either a Teddy or a Game event (adjacently placed in two unique boxes) when returning to the lab one week later. Naïve coders registered looking time towards the two dissimilar looking boxes 10 s preceding spontaneous recall. The children having spontaneous recollections looked reliably longer at their Target box (relative to Foil), whereas this was not the case for a group of matched (by gender, event, and CDI) controls producing no spontaneous memories. The findings are important for understanding how visual attention facilitates spontaneous recall.

Inhalation anesthesia and shielding devices to allow accurate preclinical irradiation of mice with clinical linac-based systems: Design and dosimetric characteristics

This technical note describes two devices to enable accurate irradiation of mice on clinical linac-based systems. To study the effects of radiation in murine, preclinical animal models, controlled and accurate dosing is important. This is not only important when specific volumes need to be irradiated, but also when the whole animal body is irradiated. To enable both purposes, we designed two devices. One device to administer Total Body Irradiation (TBI) simultaneously to six, free walking mice, and a second device, denoted as target box, in which we irradiate specific parts of the mice whilst organs-at-risk (OAR) are protected. In this latter device, we can position the mice in multiple ways. One configuration allows to sedate twelve mice simultaneously by isoflurane inhalation anesthesia and protect the body by lead shielding to allow radiation of the head only. Alternatively, the target box can be used to sedate maximal 4 mice simultaneously to irradiate the flank or paws only. All these setups allow high experimental throughput and thus a minimal occupation of the clinical equipment. As measured, the delivered radiation dosages in the regions of interest were accurate for both devices. In this technical note, we describe the design and build of these devices.

Surface connection between the Ionian Sea and different areas of the Mediterranean derived from drifter data

The surface connection between the Ionian Sea (central Mediterranean Sea) and the surrounding areas is studied by looking at the statistical properties of 1632 near-surface Lagrangian trajectories. The choice of the area is due to the key role in the dynamics of the Mediterranean Sea and to the geographical distribution of data. The Lagrangian drifter data were taken from the OGS Mediterranean drifter database, which gathers drifter data collected in the Mediterranean Sea from various institutions and countries between 1986 and 2016. The database has proved to be sufficiently complete, but the spatial and temporal data coverage are less satisfactory in the case of selection of a specific study area and temporal range. The strategy used in this work aims to limiting the problem of data coverage by choosing many target boxes around the study area, choice based on drifter trajectories, current patterns and areas of interest. The pseudo-eulerian analysis obtained from all trajectories passing through the Ionian Sea show the main dynamic structure present in the central Mediterranean Sea (e.g. Atlantic Ionian Stream, Mid Ionian Jet, etc.).On the other hand, the data density decreases progressively toward the eastern and western sectors of the Mediterranean. The highest connection was observed with the Strait of Sicily, Eastern Ionian and Adriatic Sea, with connection percentages 30%, 25% and 16% respectively. The transit times between the Ionian Sea and these target boxes are about 20–50 days. The Ionian Sea is characterized by phenomena of inversion of the basin surface circulation, from cyclonic to anti-cyclonic, over a ten-year timescale, the so-called Adriatic-Ionian Bimodal Oscillating System (BiOS, e.g. Gačić et al., 2010; Gačić et al., 2011). An application of the “target box methodology” has been used to describe how the cyclic variability of the dynamic surface currents generates an equally cyclical fluctuation of the connectivity between the Ionian Sea and the surrounding areas during the BiOS phases.

Effect of Visual Experience on Spatial Learning of Rats

Background: Visual information acquired through observation plays a pivotal role in learning a movement pattern and motor control. Objectives: This study aimed to analyze the effect of visual deprivation on learning in rats. Methods: The study sample consisted of 12 male rats, divided into three groups. Group I consisted of four rats and was considered the control group. Transection of the two optic nerves was performed on rats of group II on the seventh day after birth and group III on the seventh week after birth to develop the early blind and late blind models, respectively. A T-shaped maze device was utilized to evaluate the learning behavior of rats. Rats of groups II and III were blinded by the optic nerve surgery. In total, 20 trials per day were conducted for nine consecutive days, in which the time and number of correct arm entries were recorded. The ANOVA and Kruskal-Wallis tests were employed to analyze the results in SPSS software (version 16.0). A p-value of less than 0.05 was considered statistically significant. Results: There was a significant difference between the three groups regarding the number of correct arm entries on days one, three, and seven (P&lt;0.05). There was also a significant difference between the three groups regarding how long it took them to enter the target box on days two and three (P&lt;0.05). Group III had a lower number of entries to the target box and it took them a significantly longer time to enter the target box, compared to the other groups. Conclusion: According to the results, visual deprivation may affect the learning of rats during the early days; however, their learning levels increased over the following days. Moreover, the early blinded rats had a higher level of learning than the lately blinded adult rats and the same level of learning as that of the sighted adult rats.

Adaptive Period Embedding for Representing Oriented Objects in Aerial Images

We propose a novel method for representing oriented objects in aerial images named adaptive period embedding (APE). Although traditional object detection methods represent objects using horizontal bounding boxes, the objects in aerial images are oriented. Calculating the angle of the object is a yet challenging task. Almost all previous object detectors for aerial images directly regress the angle of objects, they use complex rules to calculate the angle, and their performance is limited by the rule design. In contrast, our method is based on the angular periodicity of oriented objects. The angle is represented by two 2-D periodic vectors the periods of which are different, so the vector is continuous as the shape changes. The label generation rule is simpler and more reasonable compared with previous methods. The proposed method is general and can be applied to other oriented detector. Besides, we propose a novel intersection over union (IoU) calculation method for long objects named length-independent IoU (LIIoU). We intercept part of the long side of the target box to get the maximum IoU between the proposed box and intercepted target box. Thereby, some long boxes will have corresponding positive samples. Our method reaches the first place of DOAI2019 competition task1 (oriented object) held in a workshop on detecting objects in aerial images in conjunction with IEEE CVPR 2019.

Automatic Detection and Segmentation on Gas Plumes from Multibeam Water Column Images

The detection of gas plumes from multibeam water column (MWC) data is the most direct way to discover gas hydrate reservoirs, but current methods often have low reliability, leading to inefficient detections. Therefore, this paper proposes an automatic method for gas plume detection and segmentation by analyzing the characteristics of gas plumes in MWC images. This method is based on the AdaBoost cascade classifier, combining the Haar-like feature and Local Binary Patterns (LBP) feature. After obtaining the detected result from the above algorithm, a target localization algorithm, based on a histogram similarity calculation, is given to exactly localize the detected target boxes, by considering the differences in gas plume and background noise in the backscatter strength. On this basis, a real-time segmentation method is put forward to get the size of the detected gas plumes, by integration of the image intersection and subtraction operation. Through the shallow-water and deep-water experiment verification, the detection accuracy of this method reaches 95.8%, the precision reaches 99.35% and the recall rate reaches 82.7%. Integrated with principles and experiments, the performance of the proposed method is analyzed and discussed, and finally some conclusions are drawn.

Finger orthoses for management of joint hypermobility disorders: Relative effects on hand function and cognitive load.

Background:Joint hypermobility refers to joints that move beyond their normal limits. Individuals with hypermobility of the fingers experience difficulties in activities of daily living. Finger orthoses are available for managing hypermobility of the fingers, but their effectiveness has received little attention in scholarly literature.Objectives:To determine if use of custom fit finger orthoses leads to improvements in time needed to perform standardised hand function tests, and attentional demand required to perform these tests, in individuals with joint hypermobility syndrome, Hypermobile Ehlers-Danlos syndrome or Classical Ehlers-Danlos syndrome.Study design:Repeated-measures study.Methods:Fourteen participants performed three different hand function tests (target box and block test, writing and picking up coins), with and without their finger orthoses. Time to complete each test was recorded as a measure of functional performance. Brain activity was recorded in the pre-frontal cortices as a measure of attentional demand.Results:Functional performance significantly improved for all but one test (picking up coins with non-dominant hand) when participants wore finger orthoses (p < 0.05). Activity in the pre-frontal cortex was lower when using the orthosis to perform the coin test (dominant hand; p < 0.05). No differences were observed in other tests (p > 0.05).Conclusions:Results suggested that finger orthoses improved hand function and provided limited evidence to suggest that they may also affect attentional demand. While the limited sample does not provide conclusive evidence supporting the use of finger orthosis in this clinical population, results warrant further investigation in large scale longitudinal studies or randomised controlled trials.

Children With Developmental Coordination Disorder Exhibit Greater Stepping Error Despite Similar Gaze Patterns and State Anxiety Levels to Their Typically Developing Peers.

This study examined stepping accuracy, gaze behavior, and state-anxiety in children with (N = 21, age M = 10.81, SD = 1.89) and without (N = 18, age M = 11.39, SD = 2.06) developmental coordination disorder (DCD) during an adaptive locomotion task. Participants walked at a self-selected pace along a pathway, placing their foot into a raised rectangular floor-based target box followed by either no obstacles, one obstacle, or two obstacles. Stepping kinematics and accuracy were determined using three-dimensional motion capture, whilst gaze was determined using mobile eye-tracking equipment. The children with DCD displayed greater foot placement error and variability when placing their foot within the target box and were more likely to make contact with its edges than their typically developing (TD) peers. The DCD group also displayed greater variability in the length and width of their steps in the approach to the target box. No differences were observed between groups in any of the gaze variables measured, in mediolateral velocity of the center of mass during the swing phase into the target box, or in the levels of self-reported state-anxiety experienced prior to facing each task. We therefore provide the first quantifiable evidence that deficits to foot placement accuracy and precision may be partially responsible for the increased incidence of trips and falls in DCD, and that these deficits are likely to occur independently from gaze behavior and state-anxiety.