2020년도 인터넷정보학회 추계학술발표대회
Current Result Document : 3 / 3
| 한글제목(Korean Title) |
Covid-19 질병의 설명 가능한 분류를 위한 포인트 깊이주의 유도 네트워크 |
| 영문제목(English Title) |
Point Depth Attention Guided Network For Explicable Classification Of Covid-19 Disease |
| 저자(Author) |
윤지영
한명묵
Mainak-Basak
Jiyoung Yun
Myung-Mook Han
|
| 원문수록처(Citation) |
VOL 21 NO. 02 PP. 0093 ~ 0094 (2020. 10) |
한글내용
(Korean Abstract) |
|
영문내용
(English Abstract) |
With the of demand for the screening of millions of prospective COVID-19 cases and the high false negative rate in the PCR tests, the need to use radiological images (like X-Ray) to test an alternative easy screening method is of great importance. Automated screening of COVID-19 from chest X-ray scan is not only urgent but also relevant worldwide during the "Novel Coronavirus" outbreak. In this paper, we report our attempt to achieve a highly detailed classification of COVID-19 disease from weak-label chest X-ray scans. Using a dense attention-guided mechanism in the convolution neural network, we implement a new approach to countering the data inadequacy issue. Classic CNN, however, has the issue of consuming processing power. Thus, instead of the standard CNN model, dilated CNN model is used to tackle over-parameterization and feature loss problem. This reduces the training time and improves the accuracy, decreasing the total overhead computation.
|
| 키워드(Keyword) |
|
| 파일첨부 |
PDF 다운로드
|
