MediaPipe with LSTM Architecture for Real-Time Hand Gesture Recognization

被引：1

作者：

Biswas, Sougatamoy ^{[1
]}

Nandy, Anup ^{[1
]}

Naskar, Asim Kumar ^{[2
]}

Saw, Rahul ^{[1
]}

机构：

[1] Natl Inst Technol Rourkela, Comp Sci & Engn, Rourkela, Odisha, India

[2] Natl Inst Technol Rourkela, Elect Engn, Rourkela, Odisha, India

来源：

COMPUTER VISION AND IMAGE PROCESSING, CVIP 2023, PT II | 2024年 / 2010卷

关键词：

Gesture recognition; LSTM; MediaPipe; Human-computer interaction; MODEL;

D O I：

10.1007/978-3-031-58174-8_36

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

Gesture recognition plays a vital role in the area of research for human-computer interaction (HCI). The integration of MediaPipe with Long Short Term Memory (LSTM) architecture holds tremendous potential for real-time hand gesture recognition. MediaPipe provides a robust and versatile framework for capturing and processing multimedia input, such as video streams from cameras or pre-recorded video files. The temporal modeling capabilities of LSTM captures the temporal dynamics of hand gestures. This research paper aims to present a novel method utilizing the MediaPipe with LSTM architecture for real-time hand gesture recognition. A test on real-time gesture recognition is performed to evaluate the performance of the suggested model. Our results demonstrate that the suggested method outperforms other state-of-the-art approaches on our custom made dataset with an accuracy of 98.99%.

引用

页码：422 / 431

页数：10

共 17 条

[1] Incorporating Relative Position Information in Transformer-Based Sign Language Recognition and Translation [J].