Logo image
Multi-loss Convolutional Networks for Semantic Segmentation
Thesis

Multi-loss Convolutional Networks for Semantic Segmentation

Lin, Che-Wei
Masters, 國立清華大學, 資訊工程學系
2014

Abstract

卷積類神經網路 影像內容分割 多重損失評估 Convolutional Neural Network Semantic Segmentation Multi-loss
This thesis presents a semantic segmentation method based on fully-convolutional network (FCN). We focus on increasing mean-class accuracy by adding other steps that help FCN to find more small objects: i) modulating the dropout rates, ii) combining multiple loss functions, and iii) expanding small object areas. Our approach shows that the above steps can significantly increase mean-class accuracy without sacrifice too much per-pixel accuracy. We also provide experimental observations on the relationship between the area-expanding method and the CNN model. Finally, we discuss how to improve the workflow and what we have learned from the experiments of training with multi-loss functions.

Metrics

1 Record Views

Details

Logo image