Indian Sign Language Gesture Recognition using Image Processing and Deep Learning

Bhagat, Neel Kamal; Vishnusai, Y; Rathna, G. N.

doi:10.1109/dicta47822.2019.8945850

Cited by 66 publications

(12 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Also, Bhagat et al [15] used a convolutional kernel with LSTMs for training the videos over a 3DCNN architecture. They proposed two Convolutional LSTM based architectures, one to train the depth and RGB videos separately and another to train them concurrently using dual-channel architecture.…”

Section: Recognition Methodsmentioning

confidence: 99%

“…Words (40) 3DCNN [13] Words (10) ConvLSTM [15] Words (25) 3DCNN [12] Words (20) 2DCNN [2] Words (30) LSTM [14] Based on comparison with Table 5, our system consists of dataset videos (35), image capturing (vision-based) and type of image (dynamic). With this number of gestures, the model outperforms previous models that are based on convLSTM method.…”

Section: Recognition Methods Reference Numbermentioning

confidence: 99%

“…Letters (32) 25,600 image Arabic [5] Numbers (0-9) 20000 image Bhutanese [6] Letters (26) 61614 image American [7] Letters (23), numbers (0-10), words (67) 35,000 image English [8] Word (15) 13,500 image American [9] Word(20) 6600 video Italian [2] Letters (30) and numbers (1-5) 1400 video German [10] Letters (H -J) and words (8) 300 video Brazilian [11] Words (25) 200 video Arabic [12] Word(40) 8000 video Arabic [13] Words (30) 1500 video Argentinian [14] Words (10) 1080 video Indian [15] Letters (26) 34627 image American [16] Words (5) 500 video Thai [17] Paper-Dynamic Sign Language Recognition Based on Real-Time Videos…”

Section: Size Language Refmentioning

confidence: 99%

“…ConvLSTM have proven to be successful in sequential images recognition [15]. The ConvLSTM is a gathering Convolution and LSTM.…”

Section: The Proposed Modelmentioning

confidence: 99%

See 3 more Smart Citations

Dynamic Sign Language Recognition Based on Real-Time Videos

et al. 2022

View full text Add to dashboard Cite

Sign language is the main communication tool for the deaf and hard of hearing. Deaf people cannot interact with others without a sign language interpreter. Accordingly, sign language recognition automation has become an important application in artificial intelligence and deep learning. Specifically, the recognition of Arabic sign language has been studied using many smart and traditional methods. This research provides a system to recognize dynamic Saudi sign language based on real time videos to solve this problem. We constructed a dataset for Saudi sign language in terms of videos in the proposed system. The dataset was then used to train a deep learning model using convolutional long short-term memory (convLSTM) to recognize the dynamic signs. Implementing such a system provides a platform for deaf people to interact with the rest of the world without an interpreter to reduce deaf isolation in society.

show abstract

Section: Recognition Methodsmentioning

confidence: 99%

Section: Recognition Methods Reference Numbermentioning

confidence: 99%

Section: Size Language Refmentioning

confidence: 99%

“…ConvLSTM have proven to be successful in sequential images recognition [15]. The ConvLSTM is a gathering Convolution and LSTM.…”

Section: The Proposed Modelmentioning

confidence: 99%

See 2 more Smart Citations

Dynamic Sign Language Recognition Based on Real-Time Videos

et al. 2022

View full text Add to dashboard Cite

show abstract

“…Neel Kamal Bhagat et al proposed a new model (RGB-D) that accomplishes mapping between the depth and the RGB pixels and various models were utilized for preparing where the depth segmented static model accomplishes accuracy of 98.81 % [5].…”

Section: Computer Vision Based Systemsmentioning

confidence: 99%

Indian Sign Language Recognition Using Canny Edge Detection

2021

IJATCSE

View full text Add to dashboard Cite

n our society, it is very difficult for hearing impaired and speech impaired people to communicate with ordinary people. They use sign languages to communicate, which use visually transmitted sign patterns, generally includes hand gestures. Sign languages being difficult to learn and non-universal, there is a barrier of communication between the hearing impaired and ordinary people. To break this barrier a system is required that can convert sign language to voice and vice versa in real-time. Here, we propose a real-time two-way system, for communication between hearing-impaired and normal people, which converts the Indian Sign Language (ISL) letters into equivalent alphabet letters and vice versa. In the proposed system, using a camera, images of ISL hand gestures are captured. Then Image pre-processing is done so that these images are ready for feature extraction. Here, a novel approach of using the Canny Edge Detection Algorithm. Once the necessary details are extracted from the image, it is matched with the data set, which is classified using Convolutional Neural Network, and the corresponding text is generated. This text is converted into a voice. Similarly, using a microphone, the voice input of an ordinary person is captured and converted into text. This text is then matched with the data set and a corresponding sign is generated. This system reduces the gap in communication between hearing-impaired and ordinary people. Our method provides 98 % accuracy for the 35 alphanumeric gestures of ISL

show abstract