Browsing by Author "Alam, Mohammad Jahangir"

Now showing 1 - 18 of 18

A Machine Learning-Based Traditional and Ensemble Technique for Predicting Breast Cancer
(Springer, 2023-05-23) Mridul, Aunik Hasan; Islam, Md. Jahidul; Asif, Asifuzzaman; Rahman, Mushfiqur; Alam, Mohammad Jahangir
Breast cancer is a physical disease and increasing in recent years. The topic is known widely in the recent world. Most women are suffering from problem of breast cancer. The disease is measured by the differences between normal and affected area ratio and the rate of uncontrolled increase of the tissue. Many studies have been conducted in the past to predict and recognize breast cancer. We have found some good opportunities to improve the technique. We propose predicting the risks and making early awareness using effective algorithm models. Our proposed method can be easily implemented in real life and is suitable for easy breast cancer predictions. The dataset was collected from Kaggle. In our model, we have implemented some different classifiers named Random Forest (RF), Logistic Regression (LR), Gradient Boosting (GB), and K-Nearest Classifier algorithms. Logistic Regression and Random Forest Classifier were performed well with 98.245% testing accuracy. Other algorithms like Gradient Boosting 91.228%, and K-Nearest 92.105% testing accuracy. We also used some different ensemble models to justify the performances. We have used Bagging LRB 94.736%, RFB 94.736%, GBB 95.614%, and KNB 92.105% accuracy, Boosting LRBO 96.491%, RFBO 99.122%, and GBBO 98.218% accuracy, and Voting algorithm LRGK with 95.614% accuracy. We have used hyper-parameter tuning in each classifier to assign the best parameters. The experimental study indicates breast cancer predictions with a higher degree of accuracy and evaluated the findings of other current studies, RFBO with 99.122% accuracy being the best performance.
A Proposition for a Low-Cost Effective Attendance Management System
(IEEE, 2020-06) Alam, Mohammad Jahangir; Faisal, Fahad; Karim, Asif
In this paper, a holistic methodology has been followed to gauge the quality of service for real-time attendance system. The experiment was done to compare the performance of an attendance server which will be cost-effective but reliable. To reduce the cost, users are connected to the biometric attendance device with a local server where the data is updated instantly. As a result, the chance of data loss is almost none. The available devices in the market are very expensive whereas our complete system may cost around USD50. On the other hand, most of the devices produce multiple data for a single user on a specific day as the data fetched from the internal memory of the respective device. But our proposed system provides a very well-ordered data for easy tracking of attendance. Moreover, our proposed framework requires a marginal power to drive the whole system. The complete system can be implemented using regular hardware and software at a very low cost.
An Approach to Detect Melanoma Skin Cancer Using Fastai CNN Models
(IEEE, 2023-05-24) Mia, Md Shazzad; Mim, Sumaiya Mustari; Alam, Mohammad Jahangir; Razib, Md.; Mahamudullah; Bilgaiyan, Saurabh
Skin cancer, which can be lethal, is essentially the improper proliferation of skin tissues. It has recently developed into one of the most dangerous sorts of additional malignancies in humans. Early detection may help the patient endure. Skin cancer is notoriously difficult to detect. Currently, computer vision performs incredibly well when used to diagnose medical images. Along with technological development and the rapid rise in computer accessibility, several machine learning algorithms and deep learning algorithms have been developed for the interpretation of medical images, especially images of skin lesions. According to our paper, there are five fast-ai CNN pre-trained models with various image pre-processing techniques that enhance the classification capability of skin lesions and make them more precise than other existing models. The HAMIOOOO dataset’s benign and malignant cancer lesions are distinguished by utilizing a number of pre-processing methods. The experimental findings showed that the suggested model improved its accuracy to 97% in both training and testing.
An Early Warning System of Heart Failure Mortality With Combined Machine Learning Methods
(Institute of Advanced Engineering and Science (IAES), 2023-08-18) Sutradhar, Ananda; Al Rafi, Mustahsin; Alam, Mohammad Jahangir; Islam, Saiful
Heart failure (HF) is currently the leading cause of morbidity and mortality worldwide. Identifying the risk of mortality at the early s tages is crucial to reducing the mortality rate. However, the traditional methods for exploring the signs of mortality are difficult and time - consuming. Whereas, m achine learning (ML) methods are superior in reducing HF’s mortality rate by providing early warnings. This study presents a novel ML classifier called imperial boost - stacked (IBS) that can serve as an effective early warning system for predicting HF mortality. Initially, we performed an efficient data balancing technique named synthetic minority oversampling technique with edited nearest neighbors ( SMOTE - ENN ) to mitigate the imbalance problem. Next, two well - known feature selection techniques , the extra tree (ET) and information gain (IG), are applied to reduce the data dimensions and select the m ost significant features. Following that, the prepared feature sets are trained with our proposed IBS classifier. Simultaneously leveraging the advantages of boosting, stacking, and multiple robust methods, it significantly correlates with the intricate pa tterns of clinical data of HF patients. Finally, the robust outcomes of 92.75% accuracy over existing studies reveal that our proposed study can effectively warn the HF mortality at early stages and reduce the burden on the healthcare sector
Compliance with Accounting Standards in Financial Reporting of Commercial Banks in Bangladesh
(University of Rajshahi, 2013) Alam, Mohammad Jahangir; Saha, Abhinaya Chandra
Financial reporting is the most effective and widely used medium through which management communicates to corporate stakeholders the operating results as well as the latest financial position of the enterprise. Users, both external and internal, depend largely on the financial information contained in the annual financial statements duly signed by external auditors in making informed judgment about an entity. At the same time, it is a tool in the hands of management to mislead the users of financial reporting. Accounting standards are the comprehensive guidelines in preparation and disclosure of financial information and serve as the benchmark for high quality financial reporting. The prime objective of this dissertation is to examine the level and extent of compliance with Bangladesh Financial Reporting Standards (BFRS/BAS) by commercial banks in Bangladesh in preparation and disclosure of financial information. The study investigates all the regulatory and professional requirements relating to the financial reporting of commercial banks and found that a complex set of banking regulation exists in Bangladesh. Some provisions of BFRS/BAS related to disclosure are in contradiction with other regulatory requirements particularly with Bangladesh Bank Circulars. It reveals that Accounting Standards are mandatory for listed companies only. The researcher prepared a disclosure checklist containing 371 items covering all the adopted BAS/BFRS in Bangladesh. Sixty annual reports of ten banks for the period of six years from 2006 to 2011 were examined under un-weighted disclosure index method to measure the compliance level. The findings reveal that the overall mean compliance score is 73.73%, implies satisfactory compliance level over the years under study. It is found that compliance score of State-owned Commercial Bank is lowest among three types of commercial banks under study. Islamic banks on the other hand, have the highest compliance score. Bank-wise analysis shows that prime Bank Limited ranked at the top and Janata Bank Limited ranked at the bottom of the banks in terms of compliance status. The study also investigates the relationship between a number of corporate attributes and extent of compliance with the requirements of accounting standards. Seven hypotheses were tested and results reveal that there exists no significant relationship between compliance score and other independent variables (except for ROA). More specifically, compliance status does not depend on the age of the bank, total assets, total capital, return on equity, earnings per share and net profit after tax (except for ROA). The dissertation also shows the results of questionnaire survey designed for the bank executives of three categories commercial banks under study. Five hypotheses were tested to find out the differences in opinions of the respondents. Total 91 variables categorized into five groups are tested to obtain desired results. Findings reveal that there exists no significant difference in opinions of the respondents regarding compliance with most of the variables of regulatory requirements, application of GAAP and qualitative characteristics of accounting information. It implies that the respondents are in agreement that commercial banks comply most of the requirements of regulatory requirements. The test results also show significant differences in opinion for most of the variables of accounting standards and application of BFRS 7. To measure the differences in opinion of the internal and external auditors, the researcher tested five hypotheses. Total 82 variables categorized into five groups are tested to obtain desired results. Findings reveal significant differences in opinion of internal and external auditors for most of the variables of accounting and auditing standards. The study finds that Lack of overall accountability and transparency, recommendatory nature of BFRS/BAS, contradiction among legal requirements, deep-rooted culture and strong mindset of the management for minimum disclosure, Lack of implementation guidelines for BFRS/BAS application, weak monitoring and supervision by regulatory bodies particularly Bangladesh Bank, lack of professional commitment of professional accountants, complex and less understandable provisions of BFRS, Intensions to give favorable audit treatments to the client and Low audit fee are some of the dominant hindrances of compliance among others. Based on the findings, some policy implications have been suggested in this dissertation which would be of immense use to practitioners and regulatory bodies as guidelines towards improving the compliance level. Disclosure checklist of BAS/BFRS prepared for this study will be useful in further compliance studies in banking sector. Users of financial reports also can take it as a guideline to understand the disclosure status of commercial banks in Bangladesh.
Dao Cao Dai: A Socio-historical Analysis of a Syncretic Vietnamese Religion and Its Relationship to Other Religions
(University of Dhaka, 2021-03-03) Alam, Mohammad Jahangir
In this dissertation Cao Dai religion is presented as one of the best examples of the outcome of Western and Asian acculturation occurred in the South of Vietnam down through the centuries. Thus, the current research makes it plain that Caodaism to a certain extent tends to be viewed as an outstanding example of a harmonious synthesis of both cultural as well as religious blends in the history of world religions. First to be considered is the fact that this research focuses Vietnamese socio-historical context with a view to gaining more comprehensive understanding of its significant role the way it played in the process of fostering the emergence of Caodaism in the South of Vietnam beginning in the early 20th century. At this point, this research work tends to explore Vietnamese social milieu in order to identify diverse deeply embedded cultural roots that actually served as a strong base for syncretic origin of Caodaism. As scholars view, with very few exceptions, the Vietnamese have a long established unique tradition of religious tolerance and thereby the religious amalgam is found to be very common among them. This century old tradition of religious tolerance and harmony in fact helped Vietnam to be a meeting place of religions and cultures. Therefore, more than a thousand-year old process of absorbing foreign elements and thereby its social integration has created a profound sense of Vietnamese identity. We have thus far discussed and evaluated about how Cao Dai religion emerged and it continued assimilating all borrowed elements into its own in the matrix of such a distinct Vietnamese socio-historical identity. At the same time, a further attempt is made to consider how the Cao Dai followers have kept undergoing a complex and syncretic religious life; and why the adherents never bother whether their religion is in greater peril of syncretism. In analyzing some reasons for this, present thesis explores the situations of continuous contact of Cao Dai religion with previously established traditions in Vietnam and addresses the approach how Caodaism blends and systematically internalizes borrowed elements in association with the elements of Western and other Eastern thoughts for its subsequent development. Consequently, Caodaism appears to be a new panorama and finds expression in a distinctive social system. This is, in a broad sense of the phenomenon, the unique syncretistic approach of the religion which has very smoothly incorporated, amalgamated and even adopted diverse elements locally in a sort of holistic mode. The study claims that these hybrid characteristics of the religion allow Caodaists to overcome a sense of cultural inferiority by establishing cultural equality with the West and the East. However, as the Western and Eastern traditions had been dominant streams to Caodaism, it was natural to have their impact upon the new religion. Thus, the present research also deals with a good effort in exploring philosophical understanding of Caodaism, its faiths and morals in order to assess Caodaism’s relation with the religions of Semitic and Indian origins. The conclusion drawn in relation to the study presented in this dissertation is that it may be pointed out that the central theme of Cao Dai syncretism is actually the way in which Caodaism approaches all borrowed elements from a refreshed and new viewpoint. Therefore, we can deduce from established arguments that the themes dealt with in the syncretic doctrine of Caodaism are the new versions of the themes of the older ones which have been transparently assimilated into Caodaism. It is to this comprehensive understanding that researcher hopes to contribute.
Drug Use Pattern in Upazilla Health Complexes of Bangladesh
(East West University, 12/24/2009) Alam, Mohammad Jahangir
In Bangladesh, the government healthcare system remaIns a very minor source of health care. The treatments provided by the doctors remain open to question, with instances of maltreatment or inadequate treatment. The treatments are mostly symptomatic and polypharmacy is Common, with antibiotics and vitamins prescribed widely. On the other hand, rural people sometimes do not buy all the drugs that are prescribed for them, partly because off tnancial constraint. In addition, self medication is common. The provision of drugs is the component of primary health care that patients most often demand and expect. Nevertheless, drugs continue to be in short supply, even when large portions of the health care budget are allocated for their procurement. In June 1982 Bangladesh introduced a national drug policy (NDP) and a drugs ordinance, which follow WHO guidelines on the selection of essential drugs. Since the enactment of the drug policy, the production, quality, and availability of essential drugs have significantly improved. Although consultations with doctors most commonly result in drugs being prescribed, very little is known about the proper use of drugs. The quality of health care, particularly the rational use of drugs, depends on a wide range of activities, such as making the correct diagnosis, prescribing the appropriate drug and dispensing them properly. When used rationally, drugs cure ailments; on the other hand, they may be dangerous and can threaten life when used irrationally. The aim of the current survey was to assess drug use for common diseases and to record the availability of essential drugs. The survey examined current treatment practices at outpatient clinics, including Assessment of patient care, physical examination, and the time given to each patient; assessment of the dispensing process in terms of the time taken and whether drugs were dispensed according to prescription; patient's knowledge of how to take the drugs; the availability of twelve essential drugs on the survey date; and the availability of an essential drugs list in the facilities. The drug use pattern and the quality of care were assessed in 5 Upazilla Health Complexes of Bangladesh. A total of 30 prescriptions, consultations, and drug-dispensing practices were studied, and the availability and use of essential drugs and of the essential drugs list were recorded. The average consulting time was 2.4 minute. The mean number of drugs prescribed per patient was 2.3. 33.30/0 were treated with antibiotics, and 17% with metronidazole, irrespective of the diagnoses. The availability of essential drugs was 77% and there was no essential drugs list in the health facilities. However, 73% of the drugs were prescribed by their generic names, 85% complied with the essential drugs list, and 80% were dispensed according to prescription. The average dispensing time was 29 seconds. The patient's knowledge on the drug dosage was 100%. Contents
Heart Disease Prediction Using Machine Learning
(IEEE, 2023-05-24) Bilgaiyan, Saurabh; Ayon, Tajul Islam; Khan, Aliza Ahmed; Johora, Fatema Tuj; Parvin, Masuma; Alam, Mohammad Jahangir
Mainly related to the cardiovascular system, brain, kidney, and peripheral arteries, the disease is called heart disease. Heart disease can have many causes, but high blood pressure and atherosclerosis are the main ones. Additionally, structural and physiological changes in the heart with age are largely responsible for heart disease, which can occur even in healthy individuals. They are not put to use or employed in any way. If these data were investigated and examined, diseases may be predicted or perhaps prevented. By using images of cancer cells to train a dataset, diseases like cancer may be identified and their stage can be forecasted. Similarly, to that, factors like cholesterol, diabetes, heart rate, etc. can be used to predict heart disease. It is difficult and dangerous to predict cardiac disorders. We noticed that sometimes there are multiple approaches used to solve a problem. It varies depending on the circumstances. The fact that most of the data are sparse or absent since they weren't recorded with the intention of analysis presents another difficulty. With data from four hospitals in four distinct locations, we, therefore, set out to determine which strategy would be best for forecasting the diseases. This study compares the effectiveness of various data mining methods for predicting cardiac disease, including, K-nearest neighbors, Random Forest, and Multi-layer Perceptron, Logistic Regression. The effectiveness of prediction for each approach utilized is reported after an analysis of the Data Mining methodologies. The outcome demonstrated that heart problems can be predicted with greater than 97 percent accuracy.
Hypothyroid Disease Prediction Using Machine Learning & Ensemble Methods
(2024-09-19) Zihadul Haque Prantik, Md.; Mustary Abeda, Maliha; Alam, Mohammad Jahangir
Thyroid illness is spreading more and more nowadays and it has become the most common disease that can lead to various health disorders. The main reason behind this disease is thyroid hormone disorder which can lead to problems like Hypothyroidism or Hyperthyroidism. As the problem is often underestimated, it is necessary but challenging to predict it efficiently. To diagnose Hypothyroidism, the traditional way is to perform thyroid tests based on various elements to diagnose the behavior of thyroid hormone. Machine learning algorithms play an important in detecting and predicting diseases at their early stages. Many analyses and models can be developed using machine learning classifiers to find the most effective and accurate predictions. Our study aims to use different machine-learning classification techniques and develop a model to predict hypothyroidism. We also compare machine learning methods to determine which classifier is optimal for constructing a classification model. We also aim to use ensemble learning techniques for optimal results and amplify accuracies. We have achieved 97.23% with gradient boosting and the AdaBoost classifier and gradient boosting are the best-performing models combined with the Boosting method giving the highest accuracy of 97.57% and 97.51% respectively.
Income and expenditure distribution pattern of sunamgonj haor area: implications for poverty alleviation
(BRAC University, 2004-08) Alam, Mohammad Jahangir
Fishing income (52 per cent) was the main source of income and food expenditure (85 percent) was the main sector of expenditure of the Sunamgonj haor fishermen. Absolute poor was about 48 per cent by Direct Calorie Intake (DCI) method and about 49 per cent and 60per cent fishermen (for income measurement unit) and 37 per cent and 55 per cent (for expenditure measurement unit) below the lower and upper poverty lines by Cost of Basic Need (CBN) method respectively. The poverty level of the haor fishermen for expenditure measurement unit is lower than income measurement unit. Boat and defecation facilities were significant by odds ratio and odds ratio confidence interval. Education status of the household head and ownership of cultivated land were significant only by odds ratio. Govt. and other agencies (NGO) may consider the significant factors to increase the socio-economic conditions and particularly to reduce poverty level of the haor fishermen.
IoT and ML Based Approach for Highway Monitoring and Streetlamp Controlling
(Springer Nature, 2023-06-11) Rahman, Mushfiqur; Suny, Md. Faridul Islam; Tasnim, Jerin; Zulfiker, Md. Sabab; Alam, Mohammad Jahangir; Akhund, Tajim Md. Niamat Ullah
Excessive speed and violating traffic rules may cause dangerous road accident. Some reports show that around 3700 people die every day due to road accident. Controlling vehicle speed and proper automated street lighting system may mitigate this problem. This work implements an automated internet of things and machine learning based system to control streetlamps with vehicle speed tracking. The developed machine learning model is capable to guesstimate the speed of the vehicle on the highway and report if there is any excessive speed. The automated streetlamp is integrated with the system that can provide proper illumination considering the environment condition. The proposed system showed good results after practical implementation.
Rose Plant Disease Detection using Deep Learning
(IEEE, 2023-05-24) Alvy, Md. Ali- Al; Khan, Golam Kibria; Alam, Mohammad Jahangir; Islam, Saiful; Rahman, Mokhlesur; Rahman, Mirza Shahriyar
The detection and identification of rose plant disease is the focus of this investigation. Identification and detection are essential components of contemporary agro technology. In this case, AI technology was utilized to identify a disease in rose plants, although plant disease detection is difficult for sustainable agriculture. There are several instances of rose plant disease, and as a result, fascinating decoration is being lost. Due to this situation, which is getting worse every day in Bangladesh, the economy of agricultural sector is suffering. Bangladesh's population relies heavily on agriculture industry for their revenue. This study includes some disease detection of rose plants, albeit not all plants are affected equally by the illness. The plant leaf provides the plant with vital sustenance. When a leaf is ill, the plant is at its most vulnerable. Due to the accessibility of the sick leaf, disease identification is difficult. Agriculture field must be properly assessed to see significant improvements in proposed work. The best resource for creating this kind of disease detection model is deep learning technology. Image pre-processing and model analysis are steps in the disease detection construction process. Few CNN architectures are used in this study, including ResNet50, VGG-16 (Visual Geometry Group), MobileNetV2, and Inception V3. Four diseases have been identified in rose plant leaves. Here, image processing is investigated using a discovered approach and obtain a MobileNetV2 model accuracy of 96.11%.
Skin Cancer Detection using Machine Learning Framework with Mobile Application
(IEEE, 2023-05-24) Ananna, Mariam Emam; Nayeem, Jannatul; Alam, Mohammad Jahangir; Islam, Saiful
A notable increase in skin cancer mortality, one of the most lethal kinds of cancer, has been caused by a lack of awareness of warning signals and preventative measures. The need for early skin cancer diagnosis has increased because of the fast development rate of melanoma skin cancer, its high cost of treatment, and its mortality risk. Treatment of cancer cells usually requires perseverance and manual identification. This study recommends using image synthesis and machine learning techniques to develop a system for diagnosing skin cancer. Thermoscopic pictures are the input for the pre-processing phase. Following the segmentation of the thermoscopic images, the attributes of the injured skin cells are obtained using a feature extraction approach. Utilizing a convolutional neural network classifier with deep learning, the collected characteristics are stratified. An accuracy of 89% has been discovered using the publicly available dataset.
Skin Disease Detection Employing Transfer Learning Approach- a Fine-Tune Visual Geometry Group-19
(Institute of Advanced Engineering and Science (IAES), 2023-07-15) Islam, Mohammad Al-Habib; Shahriyar, Sarkar Mohammad; Alam, Mohammad Jahangir; Kabir, Mohammad Rahmatul; Sarker, Rasel
Your skin may become damaged by skin diseases and conditions. These illnesses can cause skin changes such as rashes, inflammation, itching, and other skin changes. While some skin conditions may run in families, others may result from a person’s way of life. Skin conditions may be treated with pills, creams, ointments, changes in diet, and lifestyle modifications. Deep learning algorithms for computer vision applications have advanced quickly thanks to a significant amount of data for training the model and advancements in evaluation of proposed that can provide stronger simplifications. Undesired skin disease regions are eliminated, quality is raised, and the disease is tinted by discarding artifacts, decrease noise, and improving the image. Three augmentation techniques have raised the quantity of skin disease images. The five transfer learning models and various convolutional neural network (CNN) architectures analyzed the augmentation dataset. Visual geometry group-19 (VGG-19) offers the highest level of accuracy. Following the segmentation of the dermoscopic images, the affected skin cells' features are extracted using a feature extraction technique. The retrieved features are stratified using a CNN classifier, that is focused in deep learning. The best outcomes were obtained using the hyper-tuned VGG-19, which had test and validation accuracy of 99.21% and 99.25%, including both.
SunNet: A Deep Learning Approach to Detect Sunflower Disease
(IEEE, 2023-05-24) Sathi, Taslima Akter; Hasan, Md Abid; Alam, Mohammad Jahangir
Helianthus annuus, often known as sunflower, is a crop that is only mildly affected by drought. The agricultural sector of the economy benefits greatly from this. However, various illnesses have imposed a halt on sunflower cultivation over the world. However, many severe diseases will affect plants if corrective measures are not taken sooner. Therefore, it will have a negative impact on sunflower yield, quantity, and quality. Diagnosing a disease by hand can be a time-consuming and difficult process. Object recognition methods that use deep learning are becoming increasingly commonplace today. This study has developed a strategy for identifying diseases in sunflowers. A total of 1428 photos were utilized to complete this task. Images have also been processed using methods like resizing, adjusting contrast, and boosting color. Here, the area of the photos afflicted by the disease is segmented by using k-means clustering, and then retrieved characteristics from those regions. Four deep-learning classifiers were used to complete the classification. For the purpose of comparing classifier quality, four performance evaluation measures are computed. The best-performing classifier overall was a ResNet50 classifier, which had an average accuracy of 97.88% and the lowest accuracy is obtained from Inception V3.
Time Series Forecasting of Agricultural Products Sale Using Deep Learning
(Daffodil International University, 2023-07) Rahman, Md. Touhidur; Khatun, Eshita; Asha, Archina Mahmuda; Alam, Mohammad Jahangir
Due to the massive production of data, the time series dataset is useful for time-based prediction as it holds time related information. Different forecasting techniques help to predict in the field of the stock market, sales, healthcare, banking, weather etc. The coming out of new companies has increased every year, and they have taken part in the competition with other existing organizations with their products and services. As the competition is growing day by day, sales analysis is a must for every organization to make a strong position in the competitive market. In that case, product and area-wise sales analysis can help a company to attain its aim by accomplishing its target. The success of a business mainly relies on the sales of its product. Predicting the sales may help the company to discern the amount of growth rate. Moreover, this research target is to estimate the production amount for their next manufacture. More precise predictions can give fame and make companies successful. Though it is a challenging task, with the help of machine learning algorithms this issue can be sorted out effectively. In this research, we have analyzed the product selling rate and used Multi Step LSTM for forecasting the sales of the agricultural product for the next month sales
Vision-based Real Time Bangla Sign Language Recognition System Using MediaPipe Holistic and LSTM
(Daffodil International University, 2023-03-23) Foysol, Md. Walid; Sajal, Sk. Estiaque Ahmed; Alam, Mohammad Jahangir
The Bangla Sign Language Detection System converts the Bangla sign language into text so that deaf-mute people can communicate with ordinary people. Deaf-mute people are quite detached from society because normal individuals and deaf- mute individuals have a communication gap. On the strength of technological welfare, it is now possible to capture any deaf- mute person’s gesture, and with the help of machine learning, it can be converted into text. In this research, we adopt a development model for recognizing gestures that accommodates MediaPipe for extracting hands and posing landmarks and long short-term memory (LSTM) to train and recognize the gesture. This will convert Bangla sign language gestures into readable text. The requirements analysis served as the foundation for our proposed model, which will be carried out in four stages: collecting data and processing it, training and testing the chosen neural network, and finally, real-time testing. A gesture model is taught to recognize gestures with the help of a self-created dataset of Bangla sign language. The trained model successfully identifies the gesture, and the text equivalent of the gesture is displayed on screen. The purpose of this model is to help create a medium to communicate between normal people and the hearing impaired.
Water quality tests and behavioral factors of child diarrhoea in Dhaka slums
(BRAC University, 2007) Alam, Mohammad Jahangir
Diarrhoeal disease is one of the leading causes of morbidity and mortality in less developed countries, especially among children aged 0-5 years. It is a symptom of infection caused by a host of bacterial, viral and parasitic organisms most of which can be spread by contaminated water. Diarrhoea prevalence rate for the children in Dhaka slums is 214.29. The total costs of children's diarrhoea (adding all the direct and indirect) in Dhaka slums is Tk. 133.88 over a 15 day time interval. The water quality is measured at the point-of-use and the point-of-source by the total coliform, faecal coliform, and faecal streptococci tests per 100 ml water. The total coliform, faecal coliform, and faecal streptococci are at the point-of-source 651, 450, and 71 and at the point-of-use 919, 636, and 80 respectively. The test values at the point-of-use are greater than that at the point of- source due to drinking water contaminated by behavioral activities. Due to the almost perfect correlation between the total coliform, faecal coliform, faecal streptococci tests, we need to drop the values from these two tests (total coliform and faecal streptococci) in the econometric analysis. We will use the faecal coliform test, as it is the test most commonly referred to in the existing literature