TY - GEN
T1 - ContourRend: a segmentation method for improving contours by rendering
AU - Chen, Junwen
AU - Lu, Yi
AU - Chen, Yaran
AU - Zhao, Dongbin
AU - Pang, Zhonghua
N1 - Publisher Copyright:
© 2020, Springer Nature Switzerland AG.
PY - 2020
Y1 - 2020
N2 - A good object segmentation should contain clear contours and complete regions. However, mask-based segmentation can not handle contour features well on a coarse prediction grid, thus causing problems of blurry edges. While contour-based segmentation provides contours directly, but misses contours’ details. In order to obtain fine contours, we propose a segmentation method named ContourRend which adopts a contour renderer to refine segmentation contours. And we implement our method on a segmentation model based on graph convolutional network (GCN). For the single object segmentation task on cityscapes dataset, the GCN-based segmentation contour is used to generate a contour of a single object, then our contour renderer focuses on the pixels around the contour and predicts the category at high resolution. By rendering the contour result, our method reaches 72.41% mean intersection over union (IoU) and surpasses baseline Polygon-GCN by 1.22%.
AB - A good object segmentation should contain clear contours and complete regions. However, mask-based segmentation can not handle contour features well on a coarse prediction grid, thus causing problems of blurry edges. While contour-based segmentation provides contours directly, but misses contours’ details. In order to obtain fine contours, we propose a segmentation method named ContourRend which adopts a contour renderer to refine segmentation contours. And we implement our method on a segmentation model based on graph convolutional network (GCN). For the single object segmentation task on cityscapes dataset, the GCN-based segmentation contour is used to generate a contour of a single object, then our contour renderer focuses on the pixels around the contour and predicts the category at high resolution. By rendering the contour result, our method reaches 72.41% mean intersection over union (IoU) and surpasses baseline Polygon-GCN by 1.22%.
KW - Contour renderer
KW - Convolution neural networks
KW - Graph convolutional network
KW - Image segmentation
UR - http://www.scopus.com/inward/record.url?scp=85097651934&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-64221-1_22
DO - 10.1007/978-3-030-64221-1_22
M3 - Conference Proceeding
AN - SCOPUS:85097651934
SN - 9783030642204
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 251
EP - 260
BT - Advances in Neural Networks – ISNN 2020 - 17th International Symposium on Neural Networks, ISNN 2020, Proceedings
A2 - Han, Min
A2 - Qin, Sitian
A2 - Zhang, Nian
PB - Springer Science and Business Media Deutschland GmbH
T2 - 17th International Symposium on Neural Networks, ISNN 2020
Y2 - 4 December 2020 through 6 December 2020
ER -