Fusion and Sense Making of Heterogeneous Sensor Network and Other Sources
Abstract
PI Team has proposed to utilize an adaptive multimodal fusion framework that uses both training data and web resources for scene classification, the experimental results on the benchmark datasets show that the proposed text-aided scene classification framework could significantly improve classification performance. Experimental results also show that the adaptive multimodal fusion mechanism can effectively individualize reliability-dependent weighting modulation for every new observation.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 16, 2017
- Accession Number
- AD1032112
Entities
People
- Kezhi Mao
- Zhongxiang Shen
Organizations
- Nanyang Technological University