Department of Computer Science and Software Engineering-Scopes

Permanent URI for this collectionhttps://rda.sliit.lk/handle/123456789/2228

Browse

Now showing 1 - 11 of 11

Embargo
AI - Driven Smart Bin for Waste Management
(2020 2nd International Conference on Advancements in Computing (ICAC), SLIIT, 2020-12-10) Abeygunawardhana, A. G. D. T.; Shalinda, R. M. M. M.; Bandara, W. H. M. D.; W. D. S. Anesta, D.; Kasthurirathna; Abeysiri, L.
With increasing urbanization, waste has become a major problem in the present world. Therefore, proper waste management is a must for a healthy and clean environment. Though government authorities in most countries provide various solutions for waste management, solid waste tends to make a significant impact on the environment as they do not decompose easily. This research focuses on AI (Artificial Intelligence)-driven smart waste bin that can classify the most widely available solid waste materials namely Metal, Glass, and Plastic. The smart waste bin performs the separation of waste using image processing and machine learning algorithms. The system also performs the continuous monitoring of the collected waste level by using ultrasonic sensors. A dedicated mobile application will generate the optimal routes for the available waste collectors to collect the filled bins. Moreover, with this smart bin, the challenge of recognizing each waste item is overcome by using visual data as the source. Therefore, the usage of expensive sensor devices and filtration techniques to determine the category is disregarded. The smart bin can recognize the category of solid waste, collect it to the specified container, and notify the garbage level in each container. So, it is a portable waste management system.
Embargo
Arogya -An Intelligent Ayurvedic Herb Management Platform
(IEEE, 2020-11-04) Pathiranage, N; Nilfa, N; Nithmali, M; Kumari, N; Weerasinghe, L; Weerathunga, I
Ayurvedic means a science of life and well-being with its unique approaches to social and spiritual life. Especially in Sri Lanka we have our own set of rare Ayurvedic herbs which have been utilized by generations as medicinal treatments for a variety of diseases. Absence of specialists in this area makes proper identification as well as classification of valuable herbal plants a tedious task, which is essential for better treatment. Hence, a fully automated system for herb detection and classification, information visualization regarding them is highly desirable. There are existing applications which can identify plants with low prediction accuracies, as well as to give information regarding them. However, these applications are based on foreign plant data sets that do not include valuable herbs and shrubs with medicinal qualities. Hence this research proposes an application unique to medicinal plants, which can perform all these functionalities in both online and offline approach. Here, a new Ayurvedic plant dataset prepared from scratch, and preliminary results for classification of 5 types of herbs, compared with several deep Convolutional Neural Network (CNN) models based on transfer learning are presented. Experimental results indicate Marker-based Watershed algorithm as the best object detection algorithm in a complex background, VGG-16 as the best deep CNN classification model which reached a promising testing accuracy of 99.53%, and Seq2Seq LSTM model as the best deep learning model with optimum accuracy in abstractive information summarization.
Embargo
Aspect Based Sentiment Analysis for Evaluating Movies and TV series Publisher: IEEE Cite This PDF
(2020 2nd International Conference on Advancements in Computing (ICAC), SLIIT, 2020-12-10) Cooray, T.; Perera, G.; Chandrasena, D.; Alosius, J.; Kugathasan, A.
Aspect-based sentiment analysis (ABSA) is used in different fields for analyzing customer reviews to project an overall customer opinion on certain products. With the expansion of the internet, people are provided with an inexpensive and time-saving method to express their opinion to a larger audience, while various industries are handed with the opportunity to gather free information from it to obtain market value. The implementation of machine learning methods for the evaluation of aspects related to movies and television series has not been commenced, and it could be a new development for the industry. This study focuses on conducting an ABSA on a movie or a television series based on genre, story as well as cast and crew aspects. The data collected from social media through web scraping is processed to produce adequate results to get a broad understanding on how the popularity of the movie or the television series related to above mentioned aspects. Then, each aspect is further analyzed to gather precise information belonging to each aspect. The accuracy of the results of the proposed system has been achieved over 79%. The results proved that the solution is highly successful than the former works with high business value.
Embargo
Assist: Rendering, Pipeline Management, and Pipeline Tracking Software Publisher: IEEE Cite This PDF
(2020 2nd International Conference on Advancements in Computing (ICAC), SLIIT, 2020-12-10) Salgado, M.V.I; Hettiarachchi, H.A.D.D.; Munasinghe, T.U.; Fernando, K.A.U.; Gamage, I.; Thilakarathna, T.; Cooray, N.C.
Video production is one of the most dominant industries in the 21st century, and research into the automation of tasks associated with it has drastically increased. The production of videos take place in three stages: pre-production, production, and post-production. These three stages consist of script writing, scheduling, logistics, and other administration work. There are commercial products to automate these individual tasks. Incorporating all these software into video production can be expensive and difficult to manage. This study proposes the “Assist” software to handle all processes in video production. It has resulted in a product that covers the three main stages featuring scripts, storyboards, inventory management, production progress tracking and management, and rendering. The mentioned features were designed and developed using decision tree algorithm, PyQt5, general decimation algorithm, mesh simplification algorithm, and multi-variable regression.
Open Access
Comparative Analysis of Deep Learning Models for Multi-Step Prediction of Financial Time Series
(researchgate.net, 2020-10-21) Aryal, S; Nadarajah, D; Rupasinghe, P.L; Jayawardena, C; Kasthurirathna, D
Financial time series prediction has been a key topic of interest among researchers considering the complexity of the domain and also due to its significant impact on a wide range of applications. In contrast to one-step ahead prediction, multi-step forecasting is more desirable in the industry but the task is more challenging. In recent days, advancement in deep learning has shown impressive accomplishments across various tasks including sequence learning and time series forecasting. Although most previous studies are focused on applications of deep learning models for single-step ahead prediction, multi-step financial time series forecasting has not been explored exhaustively. This paper aims at extensively evaluating the performance of various state-of-the-art deep learning models for multiple multi-steps ahead prediction horizons on real-world stock and forex markets dataset. Specifically, we focus on Long-Short Term Memory (LSTM) network and its variations, Encoder-Decoder based sequence to sequence models, Temporal Convolution Network (TCN), hybrid Exponential SmoothingRecurrent Neural Networks (ES-RNN) and Neural Basis Expansion Analysis for interpretable Time Series forecasting (N-BEATS). Experimental results show that the latest deep learning models such as NBEATS, ES-LSTM and TCN produced better results for all stock market related datasets by obtaining around 50% less Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE) scores for each prediction horizon as compared to other models. However, the conventional LSTM-based models still prove to be dominant in the forex domain by comparatively achieving around 2% less error values.
Embargo
Comparative analysis of the application of Deep Learning techniques for Forex Rate prediction
(2019 1st International Conference on Advancements in Computing (ICAC), SLIIT, 2019-12-05) Aryal, S.; Nadarajah, D.; Kasthurirathna, D.; Rupasinghe, L.; Jayawardena, C.
Forecasting the financial time series is an extensive field of study. Even though the econometric models, traditional machine learning models, artificial neural networks and deep learning models have been used to predict the financial time series, deep learning models have been recently employed to do predictions of financial time series. In this paper, three different deep learning models called Long Short-Term Memory (LSTM), Convolutional Neural Network (CNN) and Temporal Convolution Network (TCN) have been used to predict the United States Dollar (USD) to Sri Lankan Rupees (LKR) exchange rate and compared the accuracy of the models. The results indicate the superiority of CNN model over other models. We conclude that CNN based models perform best in financial time series prediction.
Embargo
Domain Specific Conversational Intelligence: Voice Based E-Channeling System
(2020 2nd International Conference on Advancements in Computing (ICAC), SLIIT, 2020-12-10) Weerathunga, W.A.H.; Lokugamage, G.N.; Hariharan, V.; Yahampath, A.D.N.H.; Kasthurirathna, D.
In this research the application of Automatic Speech Recognition, Natural Language Understanding, Neural Networks and Text To Speech Conversion is investigated to create a domain specific end to end voice based E-Channeling system. The novel idea in this research can be extended to any other domain(e.g.: Taxi Application) and build a conversational intelligence system. This system enables the user to avoid the shortcomings in the traditional doctor appointment channeling procedures. The system also have the ability to predict the doctor specialization according to the symptoms of the patient and can give emergency health tips by using the powerful Neural Network module. Domain-specific speech recognition model is created according to Sri Lankan accents and handles the context-specific to this domain(94% accuracy). Extracting the entities, handling e-channeling functions and selecting the most suitable API is done by the RASA backend. Neural Network will give the first aid and doctor specialization recommendations according to user input with a validation accuracy of 90%. Speech synthesis model will output the response in user preferred language(Sinhala, English or Tamil).
Embargo
EasyTalk: A Translator for Sri Lankan Sign Language using Machine Learning and Artificial Intelligence
(2020 2nd International Conference on Advancements in Computing (ICAC), SLIIT, 2020-12-10) Kumar, D.M.; Bavanraj, K.; Thavananthan, S.; Bastiansz, G.M.A.S.; Harshanath, S.M.B.; Alosious, J.
Sign language is used by the hearing-impaired and inarticulate community to communicate with each other. But not all Sri Lankans are aware of the sign language or verbal languages and a translation is required. The Sri Lankan Sign Language is tightly bound to the hearing-impaired and inarticulate. The paper presents EasyTalk, a sign language translator which can translate Sri Lankan Sign Language into text and audio formats as well as translate verbal language into Sri Lankan Sign Language which would benefit them to express their ideas. This is handled in four separate components. The first component, Hand Gesture Detector captures hand signs using pre-trained models. Image Classifier component classifies and translates the detected hand signs. The Text and Voice Generator component produces a text or an audio formatted output for identified hand signs. Finally, Text to Sign Converter works on converting an entered English text back into the sign language based animated images. By using these techniques, EasyTalk can detect, translate and produce relevant outputs with superior accuracy. This can result in effective and efficient communication between the community with differently-abled people and the community with normal people.
Embargo
Generating 2.5D Motion Graphics from 2D Designs
(2020 2nd International Conference on Advancements in Computing (ICAC), SLIIT, 2020-12-10) Ranawake, I.; Guruge, S.; Ahamed, A.; Kasthurirathne, D.
As of the year 2020 , the video production industry is worth 31 billion dollars in the United States alone, with more than 6000 businesses and 57000 employees, and keeps growing. The global computer animation market size is anticipated to reach USD 28.30 billion by 2025, according to Grand View Research, Inc. Such significant growth demands the tech industry to introduce better tools for making animations. In this paper, we propose our contribution, specifically for User Interface designers in the field of motion graphics. UI/UX design is found to be one of the top 5 most trending opportunities for motion designers, and our proposed system allows them to generate a 2.5D animation based on a 2D Futuristic User Interface (FUI) design. The ultimate goal is to reduce the production time of FUI animations and minimize the cost of responding to the client’s changes. Obtaining clients’ feedback directly on animations rather than still images would improve the client’s involvement in production, resulting in greater confidence and loyalty. We implement several image processing techniques such as thinning and pixel clustering for pre-processing the 2D designs to segment the given design into an array of atomic shapes. Since the thinned shapes ensure that any pixel in the design does not have adjacent pixels which are also adjacent to each other, it is possible to utilize mathematical means to approximate the shapes. Our system converts a given 2D design to a collection of animated lines and arcs distributed in 3D space that eventually can be exported to the industry-standard tool, Adobe After Effects.
Embargo
Innovative, Integrated and Interactive (3I) LMS for Learners and Trainers
(2020 2nd International Conference on Advancements in Computing (ICAC), SLIIT, 2020-12-10) Fernando, K.J.L.; Jayalath, W.J.D.L.D.D.; Ranasinghe, A.D.R.N.; Bandara, P.K.B.P.S.; De Silva, H.
3I-LMS is meant to conquer the insurmountable restraints to class/lecture room education. What are insurmountable restraints to physical classroom education and how does 3I-LMS conquer them. Firstly, the lockdown and physical distancing warranted by the spread of Covid-19. In order to overcome the restraints of the pandemic 3I-LMS would be the obvious answer. 3I-LMS, the medium for delivery and reception, is online for both learners and trainers. Secondly, ineffectiveness resulting in frail in pass rates, fast diminishing memory, faulty study techniques are overcome. 3I-LMS makes studying satisfied, gamified, engaged and enjoying. The outcome would be lasting memory, effective study techniques, higher recall rates leading to improved pass rates. Thirdly, inefficiencies such as slow access to relevant lessons, notes, slow answer evaluation, feedback, clarification to routine questions are also resolved by 3I-LMS. 3I-LMS provides for keyword search, relevant subject wise notes, instant answers for routine questions thus contributing to improved efficiencies. Additionally, 3I-LMS has unique and innovative features to assist assignment completion, emanate milestone alerts, monitoring emotions and time utilization. 3I-LMS provides for utmost security. Thus, the solution deploys face recognition and keystroke dynamics to combat impersonation, copying and unauthorized referencing.
Embargo
Smart Intelligent Troubleshooter to Solve Windows Operating System Specific Issues
(2020 2nd International Conference on Advancements in Computing (ICAC), SLIIT, 2020-12-10) Rajapakshe, D.I.K.; Shamil, M.P.P.; Paththinisekara, P.M.C.P.; Liyanage, S.K.; Samaratunge Arachchillage, U.S.S.; Kuruppu, A.
While working on computers, people frequently confront with various kinds of problems, those beyond their extensive expertise. Microsoft Windows is the widely used Operating System running on numerous personal computers and the reason which gives more irritating problems that require to be addressed. Currently, troubleshooting is considered as a costly and time-consuming approach. The SAITA is an Artificial Intelligent Troubleshooting Agent that utilizes natural language generation, machine learning, and dependency resolving and ontology-based methodologies for solving most common Windows-specific issues within a short period of time than the traditional approach. The assistant learns from the accessible data and accomplishes the task for users as performed by human experts. The main objective of this exploration venture is to distinguish the constraints of existing troubleshooting software and create an AI troubleshooting assistant to provide solutions to fix the identified user issues. The use of this assistant would be economical as an IT help desk alternative in the industry. SAITA is developed to serve as a representative troubleshooter for fundamental user issues, service issues, application issues, and perform environment setup by analyzing software. This system will be able to solve the common Windows user’s issues as same as a human with less time.