Research on Image Recognition Technology Based on Multimodal Deep Learning

arXiv:2405.03091v1 Announce Type: cross
Abstract: This project investigates the human multi-modal behavior identification algorithm utilizing deep neural networks. According to the characteristics of different modal information, different deep neural networks are used to adapt to different modal video information. Through the integration of various deep neural networks, the algorithm successfully identifies behaviors across multiple modalities. In this project, multiple cameras developed by Microsoft Kinect were used to collect corresponding bone point data based on acquiring conventional images. In this way, …

abstract adapt algorithm arxiv behavior cs.cv cs.lg deep learning human identification image image recognition image recognition technology information integration modal multi-modal multimodal multimodal deep learning networks neural networks project recognition research technology the algorithm through type video