독점 멀티 분류기의 심층 학습 모델을 사용한 약지도 시맨틱 분할 = Weakly-supervised Semantic Segmentation using Exclusive Multi-Classifier Deep Learning Model

다국어 초록 (Multilingual Abstract)

Recently, along with the recent development of deep learning technique, neural networks are achieving success in computer vision filed. Convolutional neural network have shown outstanding performance in not only for a simple image classification task, but also for tasks with high difficulty such as object segmentation and detection. However many such deep learning models are based on supervised-learning, which requires more annotation labels than image-level label. Especially image semantic segmentation model requires pixel-level annotations for training, which is very. To solve these problems, this paper proposes a weakly-supervised semantic segmentation method which requires only image level label to train network. Existing weakly-supervised learning methods have limitations in detecting only specific area of object. In this paper, on the other hand, we use multi-classifier deep learning architecture so that our model recognizes more different parts of objects. The proposed method is evaluated using VOC 2012 validation dataset.

국문 초록 (Abstract)

최근 딥러닝 기술의 발달과 함께 신경 네트워크는 컴퓨터 비전에서도 성공을 거두고 있다. 컨볼루션 신경망은단순한 영상 분류 작업뿐만 아니라 객체 분할 및 검출 등 난이도가 높은 작업에서도 탁월한 성능을 보였다. 그러나 그러한 많은 심층 학습 모델은 지도학습에 기초하고 있으며, 이는 이미지 라벨보다 주석 라벨이 더 많이 필요하다. 특히semantic segmentation 모델은 훈련을 위해 픽셀 수준의 주석을 필요로 하는데, 이는 매우 중요하다. 이 논문은 이러한 문제를 해결하기 위한 네트워크 훈련을 위해 영상 수준 라벨만 필요한 약지도 semantic segmentation 방법을 제안한다. 기존의 약지도학습 방법은 대상의 특정 영역만 탐지하는 데 한계가 있다. 반면에, 본 논문에서는 우리의 모델이사물의 더 다른 부분을 인식하도 multi-classifier 심층 학습 아키텍처를 사용한다. 제안된 방법은 VOC 2012 검증 데이터 세트를 사용하여 평가한다.

참고문헌 (Reference)

1 Durand, Thibaut, "Wildcat: Weakly supervised learning of deep convnets for image classification, pointwise localization and segmentation" 2017

2 Pinheiro, Pedro O., "Weakly supervised semantic segmentation with convolutional networks" 2 (2): 2015

3 Zhou, Yanzhao, "Weakly supervised instance segmentation using class peak response" 1 : 3791-3800, 2018

4 권동진, "The Image Segmentation Method using Adaptive Watershed Algorithm for Region Boundary Preservation" 한국인터넷방송통신학회 11 (11): 39-46, 2019

5 Kolesnikov, Alexander, "Seed, expand and constrain: Three principles for weakly-supervised image segmentation" Springer 695-711, 2016

6 Hassan Ramchoun, "Multilayer Perceptron: Architecture Optimization and Training" Universidad Internacional de La Rioja 4 (4): 26-30, 2016

7 Zhou, Bolei, "Learning deep features for discriminative localization" 1 : 2921-2929, 2016

8 Deng, Jia, "Imagenet: A large-scale hierarchical image database" 1 : 248-255, 2009

9 Neubeck, Alexander, "Efficient non-maximum suppression" 3 : 850-855, 2006

10 He, Kaiming, "Deep residual learning for image recognition" 1 : 770-778, 2016

11 Roy, Anirban, "Combining bottom-up, top-down, and smoothness cues for weakly supervised image segmentation" 2017

12 Saleh, Fatemehsadat, "Built-in foreground/background prior for weakly-supervised semantic segmentation" Springer 413-432, 2016

13 O'Shea, Keiron, "An introduction to convolutional neural networks"

14 Pan, Sinno Jialin, "A survey on transfer learning" 22 : 1345-1359, 2010