Search

In the field of computer vision, fine-grained visual categorization has attracted a lot of attention and made great progress due to convolutional neural networks and a large number of publicly available datasets. With next-generation sensing technology, RGB-D cameras can provide high-quality synchronized RGB and depth images for solving many computer vision problems. Although RGB-D cameras have been used in the context of multi-view object category detection and scene understanding, they have not been widely used in fine-grained classification. In this paper, we introduce a multiview RGB-D dataset RGBD-FG for fine-grained categorization. Currently, the dataset contains 93,051 RGB-D images covering 19 super-categories and 50 sub-categories of common vegetables and fruit, and is organized in a hierarchical manner. We provide extensive experimental results to establish state-of-the-art benchmarks for our dataset, illustrating its diversity and scope for improvement through future work. We also propose a novel modality-specific multimodal network called FS-Multimodal network, which can solve two limitations of multimodal networks trained based on fine-tuning techniques: over-fitting and lack of effective depth-specific features. We hope that our study lays the foundations for fine-grained categorization of RGB-D data.

With the rapid development of deep learning technology and other powerful tools, 3D object detection has made great progress and become one of the fastest growing field in computer vision. Many automated applications such as robotic navigation, autonomous driving, and virtual or augmented reality system require estimation of accurate 3D object location and detection. Under this requirement, many methods have been proposed to improve the performance of 3D object localization and detection. Despite recent efforts, 3D object detection is still a very challenging task due to occlusion, viewpoint variations, scale changes, and limited information in 3D scenes. In this paper, we present a comprehensive review of recent state-of-the-art approaches in 3D object detection technology. We start with some basic concepts, then describe some of the available datasets that are designed to facilitate the performance evaluation of 3D object detection algorithms. Next, we will review the state-of-the-art technologies in this area, highlighting their contributions, importance, and limitations as a guide for future research. Finally, we provide a quantitative comparison of the results of the state-of-the-art methods on the popular public datasets.

Fine-grained visual categorization (FGVC) has received a great deal of attention in recent years. Currently, several public datasets are available for FGVC. However, all these datasets were created with RGB images. RGB-D sensors can provide high-quality synchronized video in terms of both color and depth. In this paper, we introduce a multi-view, large-scale RGB-D dataset called RGBD-FG to establish a novel benchmark for FGVC in RGB-D images. RGBD-FG contains RGB data and the corresponding depth data of vegetables and fruits. Our dataset was captured by a depth sensor and contained 50 categories and a total of 93,051 RGBD images with labels, and organized in a hierarchical manner. Additionally, we used several strategies on our dataset for FGVC, including a multi-modal deep CNN. We present extensive experimental results to create state-of-the-art baselines for the dataset. We hope that this dataset can fill the gap of FGVC among the RGB-D datasets.

3D object detection in RGB-D images is a vast growing research area in computer vision. In this paper, we study the problems of amodal 3D object detection in RGB-D images and present an efficient 3D object detection system that can predict object location, size, and orientation. Unlike existing methods that either uses multistage point cloud processing or pre-computed segmentation mask to generate the 3D bounding boxes, we only leverage 2D region proposals for this task. Given a pair of color and depth image as input, we first predict 2D region proposals from the designed multimodal fusion region proposal networks and then we propose an efficient method to generate 3D bounding boxes from those region proposals by scaling down the 2D bounding boxes with a scale factor and project it to 3D space. We evaluate our system on challenging NYUv2 and SUN RGB-D dataset and compare with the state-of-the-art detection methods. The experimental results show that our method outperforms the state-of-the-art by a remarkable margin with faster detection time. We achieve the best results on the NYUv2 dataset on a 19-class object detection task while performing comparably faster detection performances on the SUN RGB-D dataset on a 10-class object detection task.

Background Diarrhea remains a major public health problem and characterization of its etiology is needed to prioritize interventions. However, most data are from single-site studies of children. We tested samples from participants of any age from 11 geographically diverse hospitals in Bangladesh to describe pathogen-specific burdens of diarrhea. Methods We utilized 2 existing diarrhea surveillance systems: a Nationwide network at 10 sentinel hospitals and at the icddr,b hospital. We tested stools from enrolled participants and nondiarrheal controls for enteropathogens using quantitative polymerase chain reaction and calculated pathogen-specific attributable fractions (AFs) of diarrhea. Results We analyzed 5516 patients with diarrhea and 735 controls. Overall, rotavirus had the highest attributable burden of diarrhea (Nationwide AF, 17.7%; 95% confidence interval [CI], 14.3–20.9%; icddr,b AF, 39.9%; 38.0–41.8%), followed by adenovirus 40/41 (Nationwide AF, 17.9%; 95% CI: 13.9–21.9%; icddr,b AF, 16.6%; 95% CI, 14.4–19.4%) and Vibrio cholerae (Nationwide AF, 10.2%; 95% CI, 9.1–11.3%; icddr,b AF, 13.3%; 95% CI: 11.9–15.1%). Rotavirus was the leading pathogen in children 5 years old, but was more geographically variable (coefficient of variation = 71.5%). Other attributable pathogens included astrovirus, norovirus, Shigella, Salmonella, ETEC, sapovirus, and typical EPEC. Conclusions Rotavirus, adenovirus 40/41, and V. cholerae were the leading etiologies of infectious diarrhea requiring hospitalization in Bangladesh. Other pathogens were important in certain age groups or sites., Patients of all ages were enrolled in the sentinel surveillance from 11 Bangladesh hospitals. Rotavirus, adenovirus 40/41, and Vibrio cholerae were the leading causes of diarrhea across age and geography with variations in frequency. Other pathogens were important in certain ages or sites.

Volume rendering methods are widely used for the high-quality visualization of various 3D datasets, especially scalar field datasets (e.g., 3D images). However, when rendering datasets with ultra-high spatial resolutions, which occupy massive (out-of-core) storage space, some traditional in-core volume rendering algorithms cannot function because the large input data can hardly be handled in the main memory. Simple modifications based on disk cache I/O do not perform well because of the overheads associated with external memory access. To solve this problem, this paper describes a semi-adaptive partitioning strategy and an efficient out-of-core visualization framework with improved volume rendering algorithms. Under this new partitioning strategy, an out-of-core dataset is spatially divided into small sub-blocks of different sizes, which are organized by a binary space partitioning (BSP) tree. Each sub-block can be loaded into the fast texture memory of the graphics hardware to be rendered by our improved volume rendering algorithms. The final result is obtained by composing the projection images of all sub-blocks after traveling the BSP tree according to the viewpoint. Experimental results indicate that the new methods are effective and efficient for visualizing out-of-core 3D scalar field datasets.

The visually impaired community suffers from an immense disadvantage in academic sector. They are also not up to speed with the staggering development of technology and portable devices. To solve this issue, the area of braille technology has been widely explored since its inception. Refreshable braille display is a small step towards the advent of smart braille devices. In this paper a single cell refreshable braille display model has been proposed that is based on micro servos, as opposed to solenoids, lead screw actuation etc., to formulate different patterns of braille symbols, each representing an alphanumerical value or a contracted word. Many existing refreshable braille displays perform outstandingly, however, it is the portability, simplicity and the cost efficiency of the proposed model that contribute to the novelty of this study. A prototype of the proposed model has been built to conduct an experiment to showcase the efficiency, cost analysis and portability analysis of the system. The results show that the presented model can be a better choice in simplified and cost effective refreshable braille display implementations.

Automatic text summarization process has been significantly explored throughout the years to cope with the staggering increase of virtual data. Text summarization process is commonly divided into two areas-Extractive and Abstractive. Extractive summarization processes largely depend on sentence extraction techniques- implementing graph models or sentence-based models. In this paper, a sentence-based model has been proposed where the sentence ranking procedure adopts fuzzy C-Means (FCM) clustering, an unsupervised classification method, for sentence extraction purpose. The sentence scoring task relies on five key features, including Topic Sentence which is the first novelty of the proposed model. Furthermore, C-Means clustering is a soft-computing technique that is usually used for pattern recognition tasks but can be improved significantly by hard clustering the membership of the elements which has not been regarded in similar processes in any of the previous works, adding to the novelty of the presented model. Standard summary evaluation techniques have been used to gauge the precision, recall and f-measure of the proposed FCM model and have been compared with different summarizers from different perspectives. The outcome shows that the FCM model surpasses the previous approaches significantly.

Nowadays, Wireless Sensor Networks (WSNs) are widely used in many areas, especially in environment applications, military applications, queue tracking, etc. WSNs are vulnerable to different types of security attacks due to various constraints such as broadcasted nature of transmission medium, deployment in open or hostile environment where they are not physically protected, less memory, and limited battery power. So, security system is the crucial requirements of these networks. One of the most notably routing attacks is the sinkhole attack where an adversary captures or insert nodes in the sensor field that advertise high quality routes to the base station. In this paper, a mechanism is proposed against sinkhole attacks which detect malicious nodes using hop counting. The main advantage of the proposed technique is that, a node can detects malicious nodes only collaborating with the neighbor nodes without requiring any negotiation with the base station. Simulation result shows that, the proposed technique successfully detects the sinkhole nodes for large sensor field.

Object recognition from RGB-D images has become a hot topic and gained a significant popularity in recent years due to its numerous applications. In this paper, we propose a novel multimodal deep convolutional neural networks architecture for RGB-D object recognition which composed of three streams with two different types of deep CNNs, where each stream can separately learn from each modality. Finally, we propose a combined architecture of joint network of these three streams to classify the objects. Compared to RGB data, RGB-D images provide additional depth information that can be represented as depth colorization methods or surface normals. Our goal is to exploit both colorization and surface normals information to encode depth images. We show that by utilizing both colorization and surface normals of depth images combined with RGB significantly can improves the classification accuracy. We evaluate our model on one of the most challenging RGB-D object dataset and achieves comparable performance to state-of-the-art methods.

Processing of meteorological satellite image data provides a wealth of information useful in earth surface and environmental applications. Particularly, it is important for the estimation of different parameters of surface energy budget. In this work, a method has been developed to estimation of hourly incoming solar radiation on the surface of Bangladesh using NOAA- AVHRR satellite digital images. The model is based on the statistical regressions between the ground truth and satellite estimated values. Hundreds of full resolution images (1.1 km) for two months of the year have been processed using ERDAS IMAGINE software. Ground solar global irradiation for one place has been estimated for two months through this application. The efficiency of this method for calculating surface insolation has been checked by estimating the relative deviation between the estimated Irradiation and measured Irradiation. The method can be used for calculation of hourly irradiation over areas in a tropical environment.

Human and computer interaction has been developed into a wide and sophisticated field. Earlier automating electrical devices were absolutely mechanical. But with the invention of computer system, many computer researchers have tried to create computer based intelligent systems to accomplish many of their functionalities. Touchscreen controlled home automation system is one of them. These fascinating efforts to create intelligent systems are to provide human being a more convenient life. Moreover, it would accelerate the working speed of users. This paper concentrates primarily on experimental experiences on home automation system with a low-cost touchscreen technology. The proposed system provide an environment in which user can give commands by touching desired position on the touchscreen to automate the control of electrical devices. The interfacing circuit is designed using electronic components available in local market to keep the cost at low level.

The purpose of the study was to ascertain the attitude of the peoples towards social forestry and to explore the relationship between the selected characteristics of the respondents and their attitude towards social forestry. Attempt was also made to find out the rank order of attitudinal statements. Local community, local leader and social forestry beneficiaries of Puthia and Charghat Upzila of Rajshahi district constituted the population of the study. A total of 182 respondents were selected as sample. Interview schedule was used for collecting data from the sampled respondents during September, 2011 to December, 2011. The four selected characteristics of the respondents were considered as the independent variables. A 5-point rating scale was used to measure the dependent variable “attitude towards social forestry”. The findings revealed that near about three-fifths (59.34%) of the respondents had moderately favourable attitude towards social forestry compared to 32.2 percent highly favourable and 8.24 percent slightly favourable attitude towards social forestry. Correlation were used to explore the relationships among the variables. Out of 4 independent variables, age and occupation of all groups of respondents had a negative relationship with their attitude forwards social forestry, whereas education and annual income did not show significant relationship. The comp uted t value indicated that there was no attitudinal significant difference among the three groups. Among the 30 statements “by improving local microclimate, social forestry can contribute significantly for the maintenance of ecological balance”, “important way to create capital”, “promotion of social forestry can bring improvement in aesthetic view of the area” and “social forestry will provide harbor to insects, pests and diseases which is detrimental for human health and agricultural production” was the 1st, 2nd, 3rd and 4th, respectively. DOI: http://dx.doi.org/10.3329/jesnr.v5i1.11585 J. Environ. Sci. & Natural Resources, 5(1): 217 - 222, 2012

The study was undertaken with a view to compare clinical and laboratory diagnoses of various bacterial diseases of poultry during the period from March 2009 to February 2010 in the laboratory of the Dept. of Microbiology and Hygiene, Bangladesh Agricultural University (BAU), Mymensingh. A total of 135 sick and dead chickens (47 broilers and 88 layers) were collected from 12 different poultry farms (4 broilers and 8 layers) of Mymensingh and Gazipur districts which were subjected for clinical followed by laboratory diagnosis. Clinical diagnosis was made on the basis of clinical history, clinical signs and postmortem findings of the affected birds whereas; confirmatory diagnosis was made by using cultural examination, Gram’s staining and different biochemical tests. In this study, out of 47 broilers, 16 (34%) chickens were clinically diagnosed as colibacillosis, 11 (23.40%) as salmonellosis and 2 (4.25%) as fowl cholera. In the same way, out of 88 layer chickens, 28 (31.82%) were diagnosed as colibacillosis, 16 (18.18%) as salmonellosis and 11 (12.50%) as fowl cholera. In laboratory, out of 47 suspected broiler chickens, 12 (25.53%) chickens were diagnosed as colibacillosis, 7 (14.89%) as salmonellosis and 0 (0%) as fowl cholera. Correspondingly of the 88 layer chickens 22 (25%) were diagnosed as colibacillosis, 11 (13.64%) as salmonellosis and 8 (9.09%) as fowl cholera. So the findings concluded that clinical diagnosis is not always accurate like laboratory diagnosis because in most cases clinical history, clinical signs and post-mortem lesions of different bacterial diseases including mixed infections are almost similar to other related diseases and it is recommended to confirm laboratory diagnosis before treatment of the diseases.DOI = http://dx.doi.org/10.3329/bjvm.v8i2.11188 Bangl. J. Vet. Med. (2010). 8 (2) : 107-115

In this paper, we propose a novel self organized sensor deployment protocol to ensure proper surveillance through better coverage in urban search and rescue (USAR) operations using customized Brownian motion scheme. Specifically, we use density control over Brownian Motion (DCBM) to make an adaptive self deployment protocol which can be used on heterogeneous wireless sensor and robot networks. Generally, Brownian Motion scatters particles in a random manner with disjoint behaviour which is not appropriate to be used as a mobility model for wireless network. Additionally, the model has higher collision rate among the moving particles. To avoid the previous two issues, every node in the network computes node density and decides particular direction at any timestamp. The nodes may take pause in their movement and start over from that location instead of starting every time from the initial starting point to make the model behave continuously. We simulated DCBM scheme on a series of different topologies and compared with Random waypoint model that shows promising insights.

The sero-prevalence of Mycoplasma gallisepticum (MG) infection of chickens in selecte d Model Breeder Poultry Farms was determined during the period January to May, 2004. To conduct this study a total of 382 sera samples were collected. Rapid Serum Plate Agglutination (SPA) test was performed using commercial MG antigen (Nobilis® MG) to detect the presence of antibodies against MG. The over all sero-prevalence of MG infection was 58.90% in the study area. The highest prevalence (62.44 %) of MG infection was found in winter season followed by summer season (53.10%). The result further revealed that the infection was higher (59.94%) in female birds than in male birds (48.57%). It was also demonstrated that the infection was higher (62.80%) in Feni sadar than in Chhagoalnaiy a thana (53.45%).

Aim: Ipriflavone conserves bone mass in postmenopausal osteoporosis. Salmon calcitonin and alendronate, two other anti-resorptive drugs, have been found to have analgesic effects in osteoporotic acute vertebral fracture. This study aimed to determine if ipriflavone also possesses such an effect. Methods: Thirty-two women with recent osteoporotic vertebral fractures were randomly assigned to ipriflavone treatment or placebo groups. Ipriflavone was given at the dose of 200 mg three times a day. Non-steroidal anti-inflammatory drugs were given ad libitum in both groups. Calcium carbonate (1 g daily) was administered to all subjects. Intensity of pain at rest, on movement and on pressure, pain rating on a 10-point visual analogue scale, degree of mobility impairment, and supplementary analgesic were assessed at the end of a 3-month period in both groups to assess the analgesic effect of ipriflavone. Results: Fourteen subjects in the ipriflavone group and 12 in the placebo group completed the trial. After 3 months, all pain variables had decreased significantly in both groups. Intensity of pain at rest and on pressure and supplementary analgesic use were significantly lower in the ipriflavone group compared to the placebo group. Conclusion: The study shows that ipriflavone has an analgesic adjuvant effect in acute osteoporotic vertebral fracture.

In this paper, we propose automated image process- ing method for mammogram image analysis to enhance the clas- sification of malignant and benign clustered micro calcifications. Mini-Mias, one of the most renowned mammogram dataset, is used for our experiment. Region of interest (ROI) is selected using connected component labeling method. Shape and textural features are computed and used in five different classifiers to validate the correctness of labeling method usage for this specific research. Experimental results show that Random Forest and Bagging classifier can produce best classification accuracy among them. We used Receiver operating characteristic (ROC) analysis to asses and differentiate the classification performance from different methods. Random Forest and Bagging, both classifiers provide more than 99% accuracy.

In this paper, we propose a novel clustering protocol for wireless sensor and robot networks (WSRN) to ensure efficient energy usage and maintain maximum connectivity among the sensors. Our algorithm mainly works in two phases: (i) creates optimal number of clusters in the region of interest via three-point centroidal voronoi diagram. (ii) aggregates data from the members of each cluster and transfers to base station. This protocol is designed using double layered adaptive clustering and unknown region exploration which can be changed when needed for specific application. Additionally, this protocol works without the prior knowledge of the deployment region. We compared the results with LEACH and LEACH-C as a proof of concept. Specifically, simulation results exploit higher level of performance improvement in terms of energy dissipation, node failure, transmission overhead, and data aggregation among large number of mobile sensors. Finally, we analyzed the protocol with different settings which manifest the viability of our design.

Research focused on salient object region in natural scenes has attracted a lot in computer vision and has widely been used in many applications like object detection and segmentation. However, an accurate focusing on the salient region, while taking photographs of the real-world scenery, is still a challenging task. In order to deal with the problem, this paper presents a novel approach based on human visual system, which works better with the usage of both background prior and compactness prior. In the proposed method, we eliminate the unsuitable boundary with a fixed threshold to optimize the image boundary selection which can provide more precise estimations. Then, the object detection, which is optimized with compactness prior, is obtained by ranking with background queries. Salient objects are generally grouped together into connected areas that have compact spatial distributions. The experimental results on three public datasets demonstrate that the precision and robustness of the proposed algorithm have been improved obviously.

Aims Paediatric Exercise tolerance test (ETT) is a well-established component in the investigation of cardiac disease in children. We aimed to investigate the current practice of ETTs amongst Paediatricians across England and Wales, and provide practice and safety data from our local experience. Methods An online survey on the current practice of ETTs was distributed and completed by 49 members of the Paediatricians with Expertise with Cardiology Special Interest Group (PECSIG) between November 2014 and March 2015 and results analysed. Simultaneously a retrospective review of all patients Results Paediatric ETTs are carried out in 23/49 (47%) of responding units. In 20/21(95%) units, the decision to do the test locally was made by a consultant, with some units having age and clinical risk criteria. 18/21 (85%) of units have Paediatric doctors supervising at least some of the tests. Of the units which did not carry out any ETTs locally, unavailability of trained professional (14/20, 70%), followed by concerns regarding safety (11/20, 55%) were the most commonly cited barriers. Over 10 years, our local data identified 53 tests carried out on patients aged 7–15 years (mean 12.6 years). 7(13%) had Congenital Heart Disease and 7(13%) had abnormal baseline ECGs. 3(5.7%) were positive for exercise related arrhythmia. An adverse event occurred in 1/53 (1.9%). Following testing, 23 (43%) patients were discharged. Conclusion ETT practice vary across DGHs in England and Wales, with availability of trained staff and safety being the main barriers to its local provision. Our local rate of adverse event was 1.9% over 10 years. Clinically significant arrhythmias were found in 5.7%. Our data support the notion that Paediatric ETTs can be safely performed in a DGH and may reduce referrals to tertiary centres. Further studies will be beneficial in formulating evidence based recommendations for Paediatric ETTs in the UK.

Background: Several vitamin A supplementation trials have failed to improve the growth rate in children. Addition of zinc to vitamin A might result in enhanced growth. Objective: This study evaluated the effect on growth in children of simultaneous supplementation with zinc and vitamin A. Design: This was a randomized, double-blind, placebo-controlled intervention trial. Six hundred fifty-three children aged 12-35 mo were randomly assigned to I of 4 intervention groups: 20 mg Zn/d d for 14 d (Z group), 60000 retinol equivalents (200000 IU) vitamin A on day 14 (A group), zinc plus vitamin A (ZA group), or placebo syrup and placebo capsule (placebo group). Weight and length were measured at enrollment and again after 3 and 6 mo. Results: Gains in weight and length during the 6-mo follow-up period were not significantly different among the 4 groups by analysis of variance. Catch-up growth also did not differ significantly among the groups. The proportions of children whose weight-for-age scores did not change or decreased were 57% in the Z group, 46% in the A group, 50% in the ZA group, and 54% in the placebo group (NS). The proportions of children whose length-for-age z scores did not change or decreased were 42% in the Z group, 48% in the A group, 53% in the ZA group, and 46% in the placebo group (NS). Conclusion: Combined short-term zinc supplementation and a single dose of vitamin A has no significant effects on weight and length increments in children over a 6-mo period.

Laparoscopic cholecystectomy (LC) has become the gold standard for the surgical treatment of gallbladder disease, but conversion to open cholecystectomy and postoperative complications are still inevitable in certain cases. Knowledge of the rate and underlying reasons for conversion and postoperative complications could help surgeons during preoperative assessment and improve the informed consent of patients. We decide to review the rate and causes of conversion and postoperative complications of our LC series. This study included 760 consecutive laparoscopic cholecystectomies from July 2006 to June 2011 at Faridpur Central Hospital and Faridpur Medical College Hospital. All patients had surgery performed by same surgeon. Conversion to open cholecystectomy required in 19 (2.5%) patients. The most common reasons for conversion were severe adhesions at calot's triangle (6, 0.83%) and acutely inflamed gallbladder (5, 0.66%). The incidence of postoperative complications was 1.58%. The most common complication was wound infection, which was seen in 5 (0.66%) patients followed by biliary leakage in 3 (0.40%) patients. Delayed complications seen in our series is port site incisional hernia (2, 0.26%). LC is the preferred method even in difficult cases. Our study emphasizes that although the rate of conversion to open surgery and complication rate are low in experienced hands, the surgeons should keep a low threshold for conversion to open surgery and it should not be taken as a step in the interest of the patient rather than be looked upon as an insult to the surgeon. Key words: Laparoscopic cholecystectomy (LC); Open cholecystectomy; Conversion; Complications; Calot's triangle DOI: http://dx.doi.org/10.3329/fmcj.v6i2.9204 FMCJ 2011; 6(2): 74-77

Laparoscopic cholecystectomy has become the gold standard treatment for symptomatic gallbladder disease. Its role in surgical treatment of acute cholecystitis has also been well defined. Here a prospective study was conducted over a 3 year period of 28 patients with acute cholecystitis at district level hospitals of Bangladesh, where many modern surgical facilities were lacking. Out of 28 patients of acute cholecystitis, 24 were operated by laparoscopic methods and rests 4 were converted to open cholecystectomy. Mean operation time was 87.95 minutes and only 2 patients had postoperative complications. This study showed the appropriate time of laparoscopic cholecystectomy for acute cholecystitis, conversion rate and complications. It may be concluded that laparoscopic cholecystectomy is feasible and beneficial to the patient with acute cholecystitis in its early phase, if necessary support and expertise is available. Key Words: Acute cholecystitis; Laparoscopic cholecystectomy; Conversion DOI: 10.3329/fmcj.v5i1.6804Faridpur Med. Coll. J. 2010;5(1):3-6