2019년 추계학술대회
Current Result Document : 2 / 2
| 한글제목(Korean Title) |
복잡한 환경에서 사용 가능한 MTCNN 모델 기반 얼굴탐지 알고리즘 구현에 관한 연구 |
| 영문제목(English Title) |
A Study on Face Detection Algorithm Implementation Based on MTCNN Model for Complex Environments |
| 저자(Author) |
부옥매
김민영
장종욱
Yumei Fu
Minyoung Kim
Jong-wook Jang
|
| 원문수록처(Citation) |
VOL 23 NO. 02 PP. 0010 ~ 0013 (2019. 10) |
한글내용
(Korean Abstract) |
|
영문내용
(English Abstract) |
Aiming at the interference of light, posture and color in the process of face detection, the accuracy of face detection has been explored and studied. The main work and innovations of this paper focus on the following aspects: Image data feature enhancement. Integrate FDDB(Face Detection Data Set and Benchmark Home), LFW(Labeled Faces in the Wild) and FaceScrub's public datasets to experiment, unifying the format, size, color and brightness of all common dataset images. Selection and optimization of neural network models. Let Tensorflow build MTCNN（Multi-task convolutional neural network model, and solve the problem of over-fitting caused by the details and noise over-learning of MTCNN training samples in the learning of the sample data, and add the Dropout layer to improve the accuracy of face detection. |
| 키워드(Keyword) |
MTCNN
Face Detection
Feature extraction
TensorFlow Convolution layer
|
| 파일첨부 |
PDF 다운로드
|
