Browsing by Author "Das, Saurav"

Now showing 1 - 4 of 4

Efficiency Demonstration of Embedding Models and Libraries for Bengali Word Vector Representation
(Daffodil International University, 23-02-12) Bulbul, Aminul Islam; Das, Saurav; Tasnim, Tamanna
Word embedding demonstrates the magic in the field of NLP Our goal is to find out the best embedding model. Finding out the best embedding model for specific tasks is difficult. Embedding demonstrates different results according to size and source of data set in various embedding tasks. The purpose of the study is to find out the performance it shows for different types of embedding tasks. Researchers have invented several embedding models after finding the magical performance of word-embedding in the field of NLP. In our paper we discussed CBOW, skip-gram and Glove models performance. The models do embedding by representing the word into vector forms. We collect 2.5 lakh Bengali newspapers articles from a renowned newspaper of BD. We trained the architectures CBOW and skip-gram, which is for wor2vec and FastText models, dataset containing 20 million Bengali words. We use the same data set for training the Glove model. For collecting such a large amount of data, we build a web scraper by using Scrapy. Gensim, FastText and the python library has been used for training these three models consequently. For evaluating the models, we perform various word embedding tasks namely word analogy, semantic and syntactic prediction of words. Surprisingly they FastText perform in a better way for semantic and syntactic tasks than others. On the other hand, for analogy task, the performance was almost same for all the models except Glove.
Enhancing Bangla video comprehension through multimodal feature integration and attention-based encoder-decoder captioning models for single-action videos
(BRAC University, 2024-05) Das, Saurav; Biswas, Shammo; Fahim, Taimoor; Sanjan, M.A.B. Siddique; Tarannum, Tasnia Alam; Alam, Md. Ashraful; Alam, Md. Golam Rabiul
Video understanding and description have an important role to play in the field of computer vision and natural language processing. The capacity of automatically generating natural language descriptions for video content has many real-world applications, for example, quoting accessibility tools up to multimedia retrieval systems. Although understanding and describing video content in natural language is a challenging job, it is more so in resource-constrained languages like Bangla. This study investigates the integration of a feature fusion method and the attention-based encoder-decoder framework to improve comprehension of videos and to generate accurate captions for single-action video clips in Bangla. We propose a novel model based on multimodal fusion by combining visual features from video frames and motion information derived from optical flow. The adopted multimodal representations are then fed into an attention-based encoder-decoder architecture aiming to generate descriptive captions in the Bangla language. To facilitate our research, we collected and annotated a new dataset comprising single-action videos sourced from various online platforms. Extensive experiments are conducted on this newly created Bangla single-action videos dataset, with the models evaluated using standard metrics like BLEU, METEOR, and CIDEr. Among the models tested, including architectural variations, the GRU-Gaussian Attention model achieves the best performance, generating captions closest to the ground truth. As this is a new dataset with no previous benchmarks, the proposed approach establishes a strong baseline for Bangla video captioning, achieving a BLEU score of 0.53 and a CIDEr score of 0.492. Additionally, we analyze the attention mechanisms to interpret the learned representations, providing insights into the model’s behavior and decision-making process. This work on developing solutions for under-resourced languages paves the way for enhanced video comprehension with potential applications in human-computer interaction, accessibility, and multimedia retrieval.
Nationality Detection Using Deep Learning
(Springer Nature, 2022-10-14) Hamim, Md. Abrar; Tahseen, Jeba; Hossain, Kazi Md Istiyak; Das, Saurav
A new intelligent monitoring model is developed by us for determining gender and nationality from frontal picture candidates utilizing the face area based on deep learning. Face recognition is influenced by a number of characteristics, including picture quality, illumination, rotation angle, blockage, and facial expression. As a result, we must first recognize an input image before converting it to a genuine input. Nationality is the most well distinguishing characteristic that is applied in every nation, and it is also important to secure authentication. Image detection is crucial in this case. Then, we can determine the facial shape, gender, and nationality of the candidate image. In the end, we return the result based on the distance comparison using the use of a library to measure the sample. There were significant discrepancies across photographs while measuring samples based on their gender and facial features. The photos used in the input must be the same as those used in the output. Picture of a frontal face with clean lighting and no blemishes at every angle of rotation. The model may be used by ordinary people, models, celebrities, actors, and others. In the end, computer can tell nationality by looking at a picture of a person’s face.
Performance Evaluation of 1kW Asynchronous and Synchronous Buck Converter-based Solar-powered Battery Charging System for Electric Vehicles
(IEEE, 2020-09-05) Das, Saurav; Haque, Md.Rezanul; Razzak, M. Abdur; Leon, Md Saiful Islam; Uddin, Mohammad Rejwan
This paper presents the design and evaluates the system performance of one-kilowatt capacity asynchronous and synchronous buck converter based solar-powered charging systems for battery-driven electric vehicles. The dc motor-operated three-wheeler rickshaw was taken for testing the systems, where a battery bank containing four series-connected sub-colloid storage type batteries of each with a capacity of 12V, 120Ah has been used. PSIM simulation software has been used to evaluate the performances of these two types of battery charging systems. Hardware prototypes of these two types of charging systems have also been made and an experimental testbed comprising a 48V battery bank of 100Ah capacity with a charging current of 6A was performed. The experimental results have also been evaluated and compared.