Qamrun Nahar Eity - Academia.edu (original) (raw)

Uploads

Papers by Qamrun Nahar Eity

Research paper thumbnail of Improvement of Face and Eye Detection Performance by Using Multi-task Cascaded Convolutional Networks

2020 IEEE Region 10 Symposium (TENSYMP), 2020

Detection of face and eyes in unrestricted conditions has been a problem for years due to various... more Detection of face and eyes in unrestricted conditions has been a problem for years due to various expressions, illumination, and color fringing. Recent studies show that deep learning methods can attain impressive performance in the identification of different objects and patterns. As various systems may use the human face as input material, the increase in facial and eye detection performance has some significance. This paper introduces an enhanced face and eye detection technique through the use of cascaded multi-task convolutional networks for our dataset. We propose in this paper a deep cascaded multi-task system that exploits their inherent correlation to improve their performance. We collected 100 videos containing about 18265 images captured from our device and applied this dataset to the process and other systems proposed. The educated model was checked on our dataset and contrasted with the Haar cascade model as well. Our proposed method achieves a 98% percent accuracy rate considering our dataset which is superior to the other techniques used to detect the face and eye from an image. Besides, this paper also reflects a study of different methods of detecting the eye and face in tabular format from videos. The experimental results however indicate that the proposed approach demonstrates enhanced eye and face detection output from videos.

Research paper thumbnail of Performance Evaluation of Bangla Word Recognition Using Different Acoustic Features

IJCSNS, 2010

... Different Acoustic Features Nusrat Jahan Lisa*1, Qamrun Nahar Eity*2, Ghulam Muhammad$ Dr. Mo... more ... Different Acoustic Features Nusrat Jahan Lisa*1, Qamrun Nahar Eity*2, Ghulam Muhammad$ Dr. Mohammad Nurul Huda#1, Prof. Dr. Chowdhury Mofizur Rahman#2 ... ICASSP'99, pp.421-424, 1999. Authors Profile: Nusrat Jahan Lisa Received B.Sc. ...

Research paper thumbnail of Neural Network on the Performance of Bangla Automatic Speech Recognition

In this paper, the performance of different Bangla (widely used as Bengali) Automatic Speech Reco... more In this paper, the performance of different Bangla (widely used as Bengali) Automatic Speech Recognition (ASR) systems based on local features (LFs) to observe the effects of multilayer neural network (MLN) on it, is evaluated. These ASR systems use 3000 sentences uttered by 30 speakers from a wide area of Bangladesh, where Bangla is used as a native language. In the experiments, at first LFs are extracted from the input speech and these LFs are inputed into a multilayer neural network (MLN) for obtaining phoneme probabilities for all the Bengali phonemes considered in this study. Then, these phoneme probabilities are modified by taking logarithm or normal values, and either of these values are inputted to the hidden Markov model (HMM) based classifier to obtain word corrrect rate (WCR), word accuracy(WA) and sentence correct rate (SCR). From this study, it is observed that the ASR method which incorporates an MLN in its arechitecture improves the word recognition accuracy with fewe...

Research paper thumbnail of Improvement of Face and Eye Detection Performance by Using Multi-task Cascaded Convolutional Networks

2020 IEEE Region 10 Symposium (TENSYMP), 2020

Detection of face and eyes in unrestricted conditions has been a problem for years due to various... more Detection of face and eyes in unrestricted conditions has been a problem for years due to various expressions, illumination , and color fringing. Recent studies show that deep learning methods can attain impressive performance in the identification of different objects and patterns. As various systems may use the human face as input material, the increase in facial and eye detection performance has some significance. This paper introduces an enhanced face and eye detection technique through the use of cascaded multi-task convolutional networks for our dataset. We propose in this paper a deep cascaded multi-task system that exploits their inherent correlation to improve their performance. We collected 100 videos containing about 18265 images captured from our device and applied this dataset to the process and other systems proposed. The educated model was checked on our dataset and contrasted with the Haar cascade model as well. Our proposed method achieves a 98% percent accuracy rat...

Research paper thumbnail of Articulatory Δ and ΔΔ parameters effect on HMM-based classifier for ASR

... Qamrun Nahar Eity Dept. of cSE ahsanullah University of Science & Technology DhaNa, Bangl... more ... Qamrun Nahar Eity Dept. of cSE ahsanullah University of Science & Technology DhaNa, Bangladesh EYmail: eityBcse@hotmail.com mohammed RoNibul alam Kotwal United International University DhaNa, Bangladesh EYmail: roNibBNotwal@yahoo.com manoj BaniN Dept. ...

Research paper thumbnail of Study of an Application Development Environment Based on Unity Game Engine

International Journal of Computer Science and Information Technology

Research paper thumbnail of Japanese phonetic feature extraction for automatic speech recognition

2010 International Conference on Signal and Image Processing, 2010

Abstract This paper presents a method for extracting distinctive phonetic features (DPFs) for aut... more Abstract This paper presents a method for extracting distinctive phonetic features (DPFs) for automatic speech recognition (ASR). The method comprises three stages: i) a acoustic feature extractor, ii) a multilayer neural network (MLN) and iii) a hidden Markov model ( ...

Research paper thumbnail of Articulatory

2010 International Conference on Computer Applications and Industrial Electronics, 2010

This paper describes an effect of articulatory Δ and ΔΔ parameters on automatic speech recognitio... more This paper describes an effect of articulatory Δ and ΔΔ parameters on automatic speech recognition (ASR). Articulatory features (AFs) or distinctive phonetic features (DPFs)-based system shows its superiority in performances over acoustic features based ASR. These performances can be further improved by incorporating articulatory dynamic parameters into it. In this paper, we have proposed such a phoneme recognition system that

Research paper thumbnail of Bangla phoneme recognition for ASR using multilayer neural network

2010 13th International Conference on Computer and Information Technology (ICCIT), 2010

... Manoj Banik , Qamrun Nahar Eity , Mohammad Nurul Huda, Ghulam Muhammad Á , Yousef AjamiAlotai... more ... Manoj Banik , Qamrun Nahar Eity , Mohammad Nurul Huda, Ghulam Muhammad Á , Yousef AjamiAlotaibi Á ... Á Department of CE, College of CIS, King Saud University, Riyadh, Kingdom of Saudi Arabia. ... consonants liNe as Table I. In the Table II, the pronunciation of /N/, /O/ and ...

Research paper thumbnail of Bangla speech recognition using two stage multilayer neural networks

Abstract This paper describes a Bangla phoneme recognition method for Automatic Speech Recognitio... more Abstract This paper describes a Bangla phoneme recognition method for Automatic Speech Recognition (ASR). The method consists of two stages: i) a multilayer neural network (MLN), which converts acoustic features, mel frequency cepstral coefficients (MFCCs), into ...

Research paper thumbnail of Improvement of Face and Eye Detection Performance by Using Multi-task Cascaded Convolutional Networks

2020 IEEE Region 10 Symposium (TENSYMP), 2020

Detection of face and eyes in unrestricted conditions has been a problem for years due to various... more Detection of face and eyes in unrestricted conditions has been a problem for years due to various expressions, illumination, and color fringing. Recent studies show that deep learning methods can attain impressive performance in the identification of different objects and patterns. As various systems may use the human face as input material, the increase in facial and eye detection performance has some significance. This paper introduces an enhanced face and eye detection technique through the use of cascaded multi-task convolutional networks for our dataset. We propose in this paper a deep cascaded multi-task system that exploits their inherent correlation to improve their performance. We collected 100 videos containing about 18265 images captured from our device and applied this dataset to the process and other systems proposed. The educated model was checked on our dataset and contrasted with the Haar cascade model as well. Our proposed method achieves a 98% percent accuracy rate considering our dataset which is superior to the other techniques used to detect the face and eye from an image. Besides, this paper also reflects a study of different methods of detecting the eye and face in tabular format from videos. The experimental results however indicate that the proposed approach demonstrates enhanced eye and face detection output from videos.

Research paper thumbnail of Performance Evaluation of Bangla Word Recognition Using Different Acoustic Features

IJCSNS, 2010

... Different Acoustic Features Nusrat Jahan Lisa*1, Qamrun Nahar Eity*2, Ghulam Muhammad$ Dr. Mo... more ... Different Acoustic Features Nusrat Jahan Lisa*1, Qamrun Nahar Eity*2, Ghulam Muhammad$ Dr. Mohammad Nurul Huda#1, Prof. Dr. Chowdhury Mofizur Rahman#2 ... ICASSP'99, pp.421-424, 1999. Authors Profile: Nusrat Jahan Lisa Received B.Sc. ...

Research paper thumbnail of Neural Network on the Performance of Bangla Automatic Speech Recognition

In this paper, the performance of different Bangla (widely used as Bengali) Automatic Speech Reco... more In this paper, the performance of different Bangla (widely used as Bengali) Automatic Speech Recognition (ASR) systems based on local features (LFs) to observe the effects of multilayer neural network (MLN) on it, is evaluated. These ASR systems use 3000 sentences uttered by 30 speakers from a wide area of Bangladesh, where Bangla is used as a native language. In the experiments, at first LFs are extracted from the input speech and these LFs are inputed into a multilayer neural network (MLN) for obtaining phoneme probabilities for all the Bengali phonemes considered in this study. Then, these phoneme probabilities are modified by taking logarithm or normal values, and either of these values are inputted to the hidden Markov model (HMM) based classifier to obtain word corrrect rate (WCR), word accuracy(WA) and sentence correct rate (SCR). From this study, it is observed that the ASR method which incorporates an MLN in its arechitecture improves the word recognition accuracy with fewe...

Research paper thumbnail of Improvement of Face and Eye Detection Performance by Using Multi-task Cascaded Convolutional Networks

2020 IEEE Region 10 Symposium (TENSYMP), 2020

Detection of face and eyes in unrestricted conditions has been a problem for years due to various... more Detection of face and eyes in unrestricted conditions has been a problem for years due to various expressions, illumination , and color fringing. Recent studies show that deep learning methods can attain impressive performance in the identification of different objects and patterns. As various systems may use the human face as input material, the increase in facial and eye detection performance has some significance. This paper introduces an enhanced face and eye detection technique through the use of cascaded multi-task convolutional networks for our dataset. We propose in this paper a deep cascaded multi-task system that exploits their inherent correlation to improve their performance. We collected 100 videos containing about 18265 images captured from our device and applied this dataset to the process and other systems proposed. The educated model was checked on our dataset and contrasted with the Haar cascade model as well. Our proposed method achieves a 98% percent accuracy rat...

Research paper thumbnail of Articulatory Δ and ΔΔ parameters effect on HMM-based classifier for ASR

... Qamrun Nahar Eity Dept. of cSE ahsanullah University of Science & Technology DhaNa, Bangl... more ... Qamrun Nahar Eity Dept. of cSE ahsanullah University of Science & Technology DhaNa, Bangladesh EYmail: eityBcse@hotmail.com mohammed RoNibul alam Kotwal United International University DhaNa, Bangladesh EYmail: roNibBNotwal@yahoo.com manoj BaniN Dept. ...

Research paper thumbnail of Study of an Application Development Environment Based on Unity Game Engine

International Journal of Computer Science and Information Technology

Research paper thumbnail of Japanese phonetic feature extraction for automatic speech recognition

2010 International Conference on Signal and Image Processing, 2010

Abstract This paper presents a method for extracting distinctive phonetic features (DPFs) for aut... more Abstract This paper presents a method for extracting distinctive phonetic features (DPFs) for automatic speech recognition (ASR). The method comprises three stages: i) a acoustic feature extractor, ii) a multilayer neural network (MLN) and iii) a hidden Markov model ( ...

Research paper thumbnail of Articulatory

2010 International Conference on Computer Applications and Industrial Electronics, 2010

This paper describes an effect of articulatory Δ and ΔΔ parameters on automatic speech recognitio... more This paper describes an effect of articulatory Δ and ΔΔ parameters on automatic speech recognition (ASR). Articulatory features (AFs) or distinctive phonetic features (DPFs)-based system shows its superiority in performances over acoustic features based ASR. These performances can be further improved by incorporating articulatory dynamic parameters into it. In this paper, we have proposed such a phoneme recognition system that

Research paper thumbnail of Bangla phoneme recognition for ASR using multilayer neural network

2010 13th International Conference on Computer and Information Technology (ICCIT), 2010

... Manoj Banik , Qamrun Nahar Eity , Mohammad Nurul Huda, Ghulam Muhammad Á , Yousef AjamiAlotai... more ... Manoj Banik , Qamrun Nahar Eity , Mohammad Nurul Huda, Ghulam Muhammad Á , Yousef AjamiAlotaibi Á ... Á Department of CE, College of CIS, King Saud University, Riyadh, Kingdom of Saudi Arabia. ... consonants liNe as Table I. In the Table II, the pronunciation of /N/, /O/ and ...

Research paper thumbnail of Bangla speech recognition using two stage multilayer neural networks

Abstract This paper describes a Bangla phoneme recognition method for Automatic Speech Recognitio... more Abstract This paper describes a Bangla phoneme recognition method for Automatic Speech Recognition (ASR). The method consists of two stages: i) a multilayer neural network (MLN), which converts acoustic features, mel frequency cepstral coefficients (MFCCs), into ...