Sign Language Recognition using VGG16 and ResNet50
Rohan Gupta, Krishnam Gupta, Chirag Pandit, Prabhjot Singh · 2024
Sign language is a vital way of carrying out the conversation among mute and deaf people. The Sign language has a flaw that it is not known to everyone. To overcome this problem, the Sign Recognition program comes into the picture. The proposed model has been planned to make conversation simpler between deaf and mute persons. For determining movements and gestures it employs a neural network along with several algorithms. The proposed model has used two major neural networks which are the Resnet50 (Residual Network) and VGG16 (Visual Geometry Group). The model has been trained based on diverse linguistics across the world. It enables user-accessible online application; identifying signs in real time and conversion of images to text are few of the key features of the proposed model. In education system enables deaf and mute people to study the content without making any extra effort. It can also be of great use in the healthcare system and could work as an effective means of communication between deaf patients and doctors, which would also improve the care quality of the patients. It is observed that the images were recognized more accurately with the help of the proposed model using both neural networks. The model was provided with sample images which were divided into two folders i.e., test and train folders among which 70 percent of images were transferred to the train folder whereas 30 percent of images were transferred to the test folder after which the model went through testing and the desired outcome of identification of gestures were achieved.