BROWSE
Detail
Conference Paper
Enhanced Trident-ResCNN 기반 음향 장면 인식 알고리즘
Cited - time in 
Share 
Share
- Issue Date
- 2021-07
- Citation
- 대한전자공학회 학술 대회 (하계) 2021, pp.1017-1020
- Publisher
- 대한전자공학회
- Language
- Korean
- Type
- Conference Paper
- Abstract
- In this paper, we propose a novel acoustic scene classification algorithm based on Trident-ResCNN to enhance a model complexity and an accuracy performance. The proposed algorithm has a smaller number of residual blocks compared to the original algorithm to reduce a model size and adopts several techniques such as CoordConv layer and S&E block to improve a classification accuracy. In an evaluation test for 15 types of acoustic scene data, the classification accuracy of proposed algorithm is improved by 0.64% despite a 74.4% reduction in a model complexity compared to Trident-ResCNN algorithm.
- KSP Keywords
- Accuracy performance, Acoustic Scene Classification, Classification algorithm, classification accuracy, model complexity