Browsing by Author "Sakib, Nazmus"

Now showing 1 - 20 of 29

3D Indoor Depth Mapping Using SIFT Feature Based ICP Registration
(Department of Computer Science and Engineering, Islamic University of Technology, Gazipur, Bangladesh, 2015-11-15) Sakib, Nazmus; Farayez, Araf
3D mapping is one of the challenging research areas of Computer Vision and Robotics. Mappings are often done with mobile robots. One of the best approach is the use of SLAM (Simultaneous Localization and Mapping) where a robot roams around and builds the map and also localizes its position on the currently built map. The robot needs the sense of depth of the environment and often equipped with depth sensor. It also needs Odometry data from the accelerometer or encoder set on the wheels. There are some other solutions where the map and localization is based on the features extracted and builds the map based on the depth images and use of loop closure by adjusting the error in perception. In this thesis work we used feature based mapping which does not need any Odometry data and the map can be built based on RGB image and Depth data. We used Kinect, a very popular depth sensor which is effective in this process under certain constraints. Here in our work we gathered the RGB and Depth data simultaneously and then chose suitable frames and found key features with SIFT algorithm. The extracted features in the RGB images has corresponding 3D co-ordinates. The 3d co-ordinates have been found through intensive experiments done on calibrating the Kinect sensor and finding the best fit value of parameters to give more accurate 2D spatial points corresponding to each frame captured. The matched 3D points were then applied in Iterative closest Point (ICP) algorithm as initial points to merge the two depth images. When two images are merged they were saved in separated storages and another frame is extracted and registered with the previous ones. In this methodology multiple frames were stitched in 3D spatial co-ordinates. After forming the indoor map the map was evaluated with respect to the lengths of the objects formed in the map and the shape of the map along different 2D planes by comparing their area. Different version of ICP give different result in time complexity and accuracy. It was found IRLS (Iteratively Reweighted Least Square) ICP gave better results. The methodology we proposed can be implemented at faster time and fewer constraints and mapping do not depend on the movement noise of the robot. So the whole process is simpler and robust and can be used in indoor mapping, object detection and security purposes.
A comparative study of consumers’ perception on real estate sector in Bangladesh
(BRAC University, 11/30/2015) Sakib, Nazmus; Islam, Mr.Tamzidul
The principal reason of real estate is to sale their property to their customers. Real estate companies are expected to support their clients with a legitimate business that matches consumers’ financial condition. Pressures of urbanization in this country are compounded by the unfavorable land man ration. Suvastu Development Limited has embarked upon a mission to make the maximum use of minimum land being sensitive to both environmental concerns and social continuity. Planned development by Suvastu Development Limited has added value to those prime areas of Dhaka City .Suvastu Development Limited has currently 300 employees, including drivers & peons. I have worked as an internship at Suvastu Development Limited from 15 August-15 November 2015. Suvastu Development Limited launched their first land project Suvastu Nazar Valley which was a milestone of the company. Suvastu Development Limited participates on REHAB housing fair two times in a year. In the fair we take a stole for selling the plots. To make the sale successful the corporate office always provides feedback to the local office. We provide the brochures, maps, & other accessories so that the sales team can easily make their sales. Finally Suvastu Development Limited has successfully established itself as one of the leading real estate developments in Bangladesh with reputation for a touch of class & dignity.
A Machine Learning Approach for Multi-Level Anxiety Screening among University-Going Students using Wireless EEG Signals
(IEEE, 2025-06-23) Sakib, Nazmus; Islam, Md Kafiul; Faruk, Tasnuva
Anxiety is a widespread mental health condition affecting millions globally, often resulting in significant emotional and physical symptoms. Accurate detection of anxiety levels is essential to provide timely interventions and prevent severe complications. This study explores a machine learning-based approach for multilevel anxiety classification among young adults using EEG signals. The GAD-7 screening tool was used to assess and categorize participants into different anxiety severity groups. EEG data was then recorded, processed, and segmented into 1, 3, and 5 second segments to evaluate the impact of segment duration on classification accuracy. Four channel combinations were tested for comparisons in performance. Feature extraction included eleven time and frequency domain features. The Bagged Trees classifier was applied to classify anxiety levels based on these features. The findings of this work show the potential of EEG-based systems as non-invasive tools for anxiety screening that could support more precise mental health diagnostics.
A secured federated learning system leveraging confidence score to identify retinal disease
(BRAC University, 2023-05) Eshan, M Sakib Osman; Nafi, Md. Naimul Huda; Sakib, Nazmus; Maruf, Md. Ahnaf Morshed; Emon, Mehedi Hasan; Reza, Tanzim; Rahman, Rafeed; Parvez, Mohammad Zavid
Federated learning is a distributed machine learning paradigm that enables multiple clients to collaboratively train a global model without sharing their local data. How- ever, federated learning is vulnerable to adversarial attacks, where malicious clients can manipulate their local updates to degrade the performance or compromise the privacy of the global model. To mitigate this problem, this paper proposes a novel method that reduces the influence of malicious clients based on their confidence. We conducted our experiments on the Retinal OCT dataset. The proposed technique significantly improves the global model’s precision, recall, F1 score, and area under the receiver operating characteristic curve (AUC-ROC). Precision rises from 0.869 to 0.906, recall rises from 0.836 to 0.889, F1 score rises from 0.852 to 0.898, and AUC-ROC rises from 0.836 to 0.889.
A Study of Cyber Bullying Classification Using Social Media and Texual Analysis Based on Machine Learning Approches
(IEEE, 2023-11-23) Aronno, Md. Shafiur Rahman; Zumma, Md.Thoufiq; Prodhan, Rashed; Zohora, Fatema Tuz; Sakib, Nazmus; Tahmiduzzaman, K.B.M.
In today's world, cyberbullying is a problem that is becoming more and more common, especially among teenagers and young people. The prevalence of social networking sites and other digital communication tools has made it simpler for offenders to harass their victims in secret and without repercussions. Natural language processing (NLP) methods have been used in recent years to categorize instances of cyberbullying and assist identify them. The language used in online conversations is examined using these approaches to look for trends and signs of cyberbullying behavior. The purpose of this study is to investigate how well NLP approaches can be used to recognize and categorize cyberbullying behavior. To provide a thorough knowledge of the many types of cyberbullying, the study will use a variety of data sources, including social media posts, chat logs, and other online conversations. Overall, this research will further our knowledge of the intricate nature of cyberbullying and shed light on the potential applications of NLP approaches to lessen its negative impacts.
Airlines Ticket Price Prediction Using Machine Learning Approach
(Daffodil International University, 22-08-12) Riaz, Ashiqur Rahman; Rahman, Dewan Sakibur; Sakib, Nazmus
The price of airline tickets is the most unstable thing nowadays. It changes abruptly during the morning and evening time. The passengers are always looking to get the tickets at the lowest price, on the other hand the sellers (Airlines) are trying to earn a huge revenue. We can see that the prices change within a short time because of some factors for which the prices are affected. There are some factors like purchasing time, fuel price, flight distance etc. The prices of the airfare depends on these factors. The passengers are not allowed to access the previous data of the flight prices to predict the best price for them but the airlines have all the information about that. In this research, we tried to find out a best model for predicting the airfare by which the passenger can get the best predicted price to travel. We have used the Random Forest regression algorithm, Decision Tree algorithm and Linear Regression Algorithm to predict the price of airline tickets. For applying the ML algorithms, we have extracted the best features from the collected data and after finishing all of the tasks we got the prediction accuracy 90.47% in Random Forest Regression, 79.20% in Decision Tree and 72.77% in Linear Regression. After all, we got the best model which is Random forest Regression Algorithm to predict the airfare price. By using this system, the customers will get a better prediction that can help them buy tickets at a lower price.
An LSTM-Based Word Prediction in Bengali
(Springer, 2022-11-14) Hasan, Mustahid; Sakib, Nazmus; Hridoy, Rashidul Hasan; Ananto, Nazmul Hossain; Akhter, Sonia; Habib, Md. Tarek
"In this paper, Bengali text information has been utilized for predicting the next word contingent based on the previous one. To do that, one should consider two key aspects such as the natural language processing (NLP) stage and the word predicting stage. When both work together, the system gets a new predicted word that is relevant to the previous word. For achieving such correct predicted words, long short-term memory (LSTM) has been used which is best known for its memory management. LSTM embeds the input words and fits them into the model, then after successful training of the model, it can predict the next word from a given sentence. The user can also initialize the number of predicted words. This paper gives an overview of word prediction for the Bengali language based on LSTM and describes the database integration and proposed approach obtained 97.60% accuracy."
Comparison of Performance of a 500 Watt Refrigeration System With One Compressor and Three Refrigerants
(Department of Mechanical and Production Engineering (MPE),Islamic University of Technology(IUT), Board Bazar, Gazipur, Bangladesh, 2012-11-15) Sakib, Nazmus; Azam, Rakib-Ul-
Currently in Bangladesh, most low capacity refrigerant systems (below 1 ton refrigerating effect) use R-12 compressors with R-12 refrigerant fluid. But with the global policy changes of phasing out R-12, these systems need to be replaced. However, changing compressors to fit the replacing fluid is a costly process. This paper finds out whether R-22 and R-134a are suitable for use with R-12 compressors and by comparative analysis determine which one is the best replacement for the systems.
Cybersecurity Awareness Assessment Based on Bangladeshi People Perspective Using Machine Learning Approaches
(IEEE, 2023-11-23) Tahmiduzzaman, K. B. M.; Zumma, Md. Thoufiq; Babi, Farzana; Sakib, Nazmus; Adnan-E-Elahe, Md.
The demands placed on cyber security continue to rise. When data is compromised from a company, organization, or individual, we realize how crucial it is to have a basic understanding of cyber security. The company should provide their employees with training on how to stay secure online. In light of the current circumstances, it is very important for everyone. Those who have less awareness on it are contributing to a rise in cybercrime. The most significant cyberattack against a developing country. As a result of this, hackers and spammers target individuals in order to execute out attacks, and Bangladesh is vulnerable to an excessive number of attacks on its online and mobile banking transactions. In essence, they apply a social engineering approach in order to get the password. In this study, we gathered data from a survey in which we asked certain questions about fundamental levels of awareness. This research has a total of 456 participants who have answered the questions. In order to get the highest level of accuracy, we employ a machine learning method.
Design and Development of a Web Base Application for Tree Nursery
(Daffodil International University, 2021-06) Sakib, Nazmus; Rahman, Tamin Nur; Ahmed, Mahfuz
Global warming and climate change refer to the increase in global average temperature. It is believed that natural events and human activities are the main causes of such a rise in global mean temperatures. Perform various activities that gradually increase the temperature. Global warming is rapidly melting our glaciers, which is extremely harmful to the planet and humans. We are talking about global warming but we are not getting any solution. Global warming has become a serious problem and requires special attention. This is not for one reason, but for many reasons, including natural and man-made. As the number of people increases, the number of lands decreases. We are not able to plant enough trees due to lack of land. The number of lands towards the city is very low, we can organize tree planting on the roof of every house. People don't know where to get good seedlings. Our website allows people to collect seedlings in a safe place. Overall, we understand that this method does not solve the problem of global warming, but it may mean that our population and society will undergo major changes.
EEG-based Mouse Cursor Control using Motor Imagery Brain-Computer Interface
(IEEE, 2024-05-03) Roja, Saima Tasfia; Bin Rafique, Sayem; Rhaman, Md. Asikur; Sakib, Nazmus; Islam, Md Kafiul
Brain-computer interface (BCI) is a system that collects, analyzes, and transforms brain signals into commands. The brain experiences repetitive, oscillatory electrical changes caused by these activities that have a very low voltage of only a few microvolts (µV).The term electroencephalogram (EEG) refers to the non-invasive recording of this electrical activity from the scalp. The signals are then analyzed in a computer to identify the desired action after signal acquisition. Relevant features are gathered and translated into commands that operate an output device or carry out the command. The user is subsequently provided with feedback to confirm that the command has been carried out correctly This work focuses on the development and implementation of a Mouse Cursor Control system using Motor Imagery (MI) BCI, using data recorded with the Emotiv EPOC+ headset and processed using our algorithm in the MATLAB software. While similar works do exist, most tend to focus on one or more aspects of data processing such as classification. We acquired the data from subjects ourselves, and after processing the data using our algorithm, the system was implemented, and the cursor was moved. This makes our system a semi-online system, as opposed to offline systems. The only limitation of our system is that the system is implemented in semi-real time. Furthermore, accuracy was tested for different frequency bands and the highest accuracy of 93.60% was achieved using the offline dataset.
Effect of artifact removal on EEG based motor imagery BCI applications
(SPIE Digital Library, 2024-01-29) Islam, Md Kafiul; Sakib, Nazmus; Anjum, Maisha
Brain computer interface (BCI) is an emerging technology where the user can establish direct communication between the electrical device and himself without any physical exertion. The EEG signal is a noninvasive and low-cost method to extract brain signal from subject. The EEG signal contains different types of information including motor sensory information originating from the motor cortex region of the brain. Research and study have shown that motor cortex generates signals similar to the signals generated during deliberate limb movements. Therefore, motor imagery (MI) signals if extracted can be utilized to operate any electrical device establishing a BCI system. However, the EEG data can contain lots of artifacts. This degrades the signal quality and also cause false positive command to the connected device. Therefore, it is crucial to remove the artifacts from the EEG signal before classification. In this project, EEG data has been collected from 12 subjects who are instructed to perform MI activity. The EEG signal is then processed and an efficient artifact removal technique has been applied. The artifact removal method applies wavelet transform theorem and artifactual probability mapping method to detect artifactual epochs and eliminate it from the signal. Useful features are then extracted from the signal and artificial neural network (ANN) classifier is applied to it. The classification accuracy has been enhanced by 15-16% on average after removal of artifacts from the EEG recordings for MI-BCI experiments. Afterwards, performance evaluation such as finding signal to noise ratio has been done to evaluate the improvement in the signal after noise removal.
Exploring the Prevalence and Triggering Factors of Migraine in University Students of Bangladesh Using Machine Learning
(Department of Electrical and Elecrtonics Engineering(EEE), Islamic University of Technology(IUT), Board Bazar, Gazipur-1704, Bangladesh, 2023-05-30) Saif, Zawwad Bin; Sakib, Nazmus; Adnan, Muhammad
Migraine is a recurring neurovascular illness causing prolonged acute pain, nausea, vomiting, and autonomic nervous system dysfunction, resulting from disrupted blood vessels and nerve signals in the brain due to unbalanced activity, whose exact cause remains unknown influencing significantly the quality of life. The study aims to explore the prevalence of migraine among Bangladesh's university students, predict their occurrence based on triggering factors using machine learning, and raise awareness to facilitate the everyday activities of migraine patients. Around 303 students from various universities in Bangladesh participated in this cross-sectional survey. in an interval between August to October of 2022 via means of a voluntarily completed online platform-based questionnaire. For the survey structure, a total of twenty factors were sorted out after keen observation that triggers the migraine and subsequently, a dataset was structured based on the factors. The prevalence of migraine and these 20 triggering factors of migraine among university students were determined through this survey. To generate a probabilistic prediction of the occurrence of migraine, nine ML algorithms have been applied for male and female participants separately considering the headache-triggering factors. With some data preprocessing and feature engineering, GridSearchCV was used to optimize the hyperparameters for each of the nine classification models to achieve more efficient results. ML algorithms are compared by examining their several performance matrices like accuracy, train score, precision, recall, F1 score, and ROC-AUC value and after extensive simulation, the Logistic Regression algorithm emerged with the highest accuracy of 78.1% for the male participants. The stacking Classifier and Random Forest Classifier emerged with the highest accuracy of 85.3% in the case of the female participants. Making use of various machine learning algorithms and clinical data in this field has the potential to make it simpler for people with migraines to identify and avoid the triggers of their condition, allowing them to go about their daily lives more comfortably
Grape Leafs Disease Detection Using Customized CNN Model
(Daffodil International University, 2024-07-13) Sakib, Nazmus
This study looks at the diagnosis of grape leaf illnesses using a Kaggle dataset that is categorized into four categories: "esca," "black rot," "healthy," and "leaf blight." The study introduces a brand-new illness categorization method based on Convolutional Neural Networks (CNN) models. For comparison, the popular pre-trained models MobileNet and VGG16 are also used. The primary goal is to offer a reliable and effective technique for the automated identification and categorization of diseases affecting grape leaves, an essential task for the timely diagnosis and medical care of illnesses in the wine industry. Preprocessing methods, such as data augmentation and normalization, are used in the study to improve model performance. Experimental assessments are performed on the dataset to compare the proposed CNN model with MobileNet and VGG16 in terms of accuracy, precision, recall, and F1-score. The modified CNN model is effective at correctly recognizing grape leaf diseases, according to the results. In summary, this thesis advances automated disease identification in viticulture by shedding light on which CNN architectures are most suited for a given job and laying the groundwork for future studies in agricultural image processing.
HR practices in Renata limited
(BRAC University, 5/23/2012) Sakib, Nazmus; Akter, Kohinur
Bangladesh is a developing country in south Asia. Its economy depends mostly on agriculture. In the earliest period jute and tea industry were very raising industry. But the scenario is changed. Now our economy mostly depends on Ready Made Garments industry. Apart from garments industry Pharmaceuticals industry is a raising industry in Bangladesh. So the scenario is changing. Now there are 225 registered pharmaceutical companies in our country. Most of them are local, but there are also some multinational and joint venture companies operating their business in our country. They are also contributing to earn foreign exchange in our country. Renata is one of the most leading pharmaceuticals company in Bangladesh. Its corporate headquarters is at plot no- 1, milk vita road, sec- 7, mirpur and it has two production sites (one is at Mirpur and another is at Rajendrapur). Modern sophisticated machineries and highly qualified and skilled professionals are the main instrument for this Renata’s success. In this report we have gone through the overall HR policies and practices of Renata Limited. Renata’s HR functions like Recruitment & Selection, Training & Development, Compensation & Benefit and Employee relation are very systematic and legal. Renata always tries to recruit efficient candidates for the vacant position. They provide management development training, overseas training, and Field forces training and development program, manager training, training program for distribution assignments and data entry operations, on the job training and off the job training, basic training and advanced training etc. Renata concentrates on employee development as well. They provide appropriate compensation and benefits like basic salary, over time, gratuity, different types of bonus etc. I have found that Renata’s HR policies and practices are systemic and sound enough but yet there are some problems like (few misunderstanding, lack of appropriate training to reduce faults). Remove misunderstanding and provide training and appropriate compensation when any accident occurs, authority should cooperate their employees to the problems. I think that is the way they should follow to improve and expand its business.
Integration of digital loyalty program for hospitality industry
(BRAC University, 2024-01) Sakib, Nazmus; Hasan, Najmul
The report describes the on-site employment experience that I obtained while working as an intern with Paragon Ceramic Industries Ltd for its sister concern Paragon Hospitality Management in the Digital Marketing department. This experience was valuable to me since it allowed me to get practical insights about the hospitality management. I have explained the obligations that were given to me and carried out by me under the guidance of the supervisor who was stationed on-site throughout the length of this report. In addition, the report offers a condensed overview of the organization's brands, in addition to a discussion of the major activities and strategic approaches that are utilized by other brands in the industry that are comparable to the organization's brands. In conclusion, the goal of the research project with the working title "Integration of digital loyalty program for hospitality industry " is to improve the quality of the interactions and services that are provided to the subscribers of the digital ecosystem in an effort to improve the overall customer experience in the digital ecosystem which will be used as the value added service for the customers of PCIL’s brands. The implications of the study have made an attempt to come up with vital elements associated to the usability and effectiveness of the platform in terms of providing a priority-based one-stop solution to the consumers.
Investigating factors influencing pedestrian crosswalk usage behavior in Dhaka city using supervised machine learning techniques
(2024-03-24) Sakib, Nazmus; Paul, Tonmoy; Ahmed, Md. Tawkir; Al Momin, Khondhaker; Barua, Saurav
Pedestrians are the most vulnerable road users and are over-represented in casualty statistics, particularly in low- and middle-income countries like Bangladesh. To ensure the safety of pedestrians, it is necessary to identify the factors underlying pedestrian behavior while crossing. Hence, this study aims to predict the pedestrian decision regarding crosswalks using supervised machine learning techniques namely, Classification and Regression Tree (CART), Random Forest (RF), and Extreme Gradient Boost (XGBoost). A questionnaire survey was conducted in twelve important locations of Dhaka, Bangladesh using 8 attributes related to crosswalk behavior. Analysis suggests RF model is the most effective in terms of prediction performances, specifically having a 96.00% F1 score and 95.83% MCC value. It has been found that unsuitability of crosswalk location, absence of guard rails on median, and inadequate lightning at night near crosswalks are the most important features for preferring to use crosswalks. The findings of the study will help policymakers and transport planners to plan accordingly in order to develop safe crosswalks.
IoT security risk analysis
(BRAC University, 2017) Sakib, Nazmus; Jerin, Ismot; Khan, Nuzhat; Quader, Shaela; Chakrabarty, Amitabha
Internet of things (IoT) has become a buzzword in today’s world to describe billions of devices, interconnected via the web. It includes a diverse range of devices, starting from wearable ultra low-powered gadgets like fitness bands to medical instruments and to home appliances to automobiles. There may be so many devices harnessing the power of IoT, however, security is still an issue for these devices as these are constrained with limited power supply, processing cycles and memory usage. The IoT sector is not impeccable, security is still a threat for the IoT devices as these devices are meant for low power usage for small scale setups. Security algorithms are not abrupt, but many of them don’t fit the IoT systems as their compatibility can only rely on products with larger form factor (Which usually means better performance, storage etc.). In such context, working with security of IoT devices has become an interesting area in computer science. Though, researchers and security professionals have developed advanced algorithms for ensuring digital security, but many of them are not suitable for the IoT world because of the restrictions we have. In our work, we tried to contribute to the matter of security in IoT devices. In this work, the main concern has been to investigate the performance of different security algorithms and compare them in terms of processing cycle and execution time in Raspberry Pi. We have worked with FLECC_IN_C and Crypto++, two different libraries with number of algorithms where we can find ecdh, ecdsa, ciphers, message authentication codes, one-way hash functions, public-key cryptosystems, key agreement schemes, and deflate compression. and measured their performance in a constrained environment. It is the first of its kind, to this work’s knowledge, to use raspberry pi which is established as black box device and implemented security algorithms on it. We implemented these libraries in different IoT platforms, showing comparisons of how these algorithms may affect a system in terms of resource utilization. The work in the end shows a summarised view of several key algorithms and decides which is better in the terms of IoT constraints.
Machine Learning Model for Computer-Aided Depression Screening among Young Adults Using Wireless EEG Headset
(Hindawi, 2023-05-31) Islam, Md Kafiul; Faruk, Tasnuva; Sakib, Nazmus
Depression is a disorder that if not treated can hamper the quality of life. EEG has shown great promise in detecting depressed individuals from depression control individuals. It overcomes the limitations of traditional questionnaire-based methods. In this study, a machine learning-based method for detecting depression among young adults using EEG data recorded by the wireless headset is proposed. For this reason, EEG data has been recorded using an Emotiv Epoc+ headset. A total of 32 young adults participated and the PHQ9 screening tool was used to identify depressed participants. Features such as skewness, kurtosis, variance, Hjorth parameters, Shannon entropy, and Log energy entropy from 1 to 5 sec data filtered at different band frequencies were applied to KNN and SVM classifiers with different kernels. At AB band (8–30 Hz) frequency, 98.43 ± 0.15% accuracy was achieved by extracting Hjorth parameters, Shannon entropy, and Log energy entropy from 5 sec samples with a 5-fold CV using a KNN classifier. And with the same features and classifier overall accuracy = 98.10 ± 0.11, NPV = 0.977, precision = 0.984, sensitivity = 0.984, specificity = 0.976, and F1 score = 0.984 was achieved after splitting the data to 70/30 ratio for training and testing with 5-fold CV. From the findings, it can be concluded that EEG data from an Emotiv headset can be used to detect depression with the proposed method.
Marketing strategy of Hotel Link Solutions: digital marketing for hotels and resorts in Bangladesh
(BRAC University, 6/22/2014) Sakib, Nazmus; Islam, Md. Tamzidul
Hotel Link Solutions is one of the leading digital marketing solution providers in today’s world. Now-a-days a website with an email address is not enough. Travelers want to check availability instantly, and then book with their booking with their credit card. They use multiple channels and mobile devices to find properties. So Hotel Link Solutions has developed a modular digital marketing solution which gives you the right tools to manage your accommodation. The marketing strategies and activities instated by Hotel Link Solutions are innovative and out of the box. The pipedrive, resonline and health check makes the whole system work like an integrated network. The business model used by Hotel Link Solutions is totally unique combination of digital marketing efforts. Hotel Link Solution Bangladesh offers the same service as the global offers. For the last one year operation in Bangladesh they have achieve considerable amount of success in the industry.