Misclassification in semantic segmentation mostly occurs in the pixels around the semantic contour. In this work, we address the task of aerial image segmentation by borrowing the kernel prior from classical edge detecting operator. We propose a module called Sobel Heuristic Kernel(SHK). Our work makes several main contributions and experimentally shows good performance. To the best of our knowledge, we are the first to combine traditional edge detection method and deep learning method in semantic segmentation. Our SHK module reaches state of the art in the Inria Aerial Image Labeling dataset.
%0 Conference Paper
%1 8451170
%A Hu, T.
%A Wang, Y.
%A Chen, Y.
%A Lu, P.
%A Wang, H.
%A Wang, G.
%B 2018 25th IEEE International Conference on Image Processing (ICIP)
%D 2018
%K (artificial Aerial Detection Heuristic Image Kernel;Kernel;Semantics;Image Labeling Segmentation;Edge contour;aerial dataset;Aerial detecting detection detection;Neural detection;image edge image intelligence);Inria method;SHK module;Sobel networks;Detectors;Convolution;Semantic operator;traditional segmentation;Image segmentation;classical segmentation;learning segmentation;semantic semantic
%P 3074-3078
%R 10.1109/ICIP.2018.8451170
%T Sobel Heuristic Kernel for Aerial Semantic Segmentation
%X Misclassification in semantic segmentation mostly occurs in the pixels around the semantic contour. In this work, we address the task of aerial image segmentation by borrowing the kernel prior from classical edge detecting operator. We propose a module called Sobel Heuristic Kernel(SHK). Our work makes several main contributions and experimentally shows good performance. To the best of our knowledge, we are the first to combine traditional edge detection method and deep learning method in semantic segmentation. Our SHK module reaches state of the art in the Inria Aerial Image Labeling dataset.
@inproceedings{8451170,
abstract = {Misclassification in semantic segmentation mostly occurs in the pixels around the semantic contour. In this work, we address the task of aerial image segmentation by borrowing the kernel prior from classical edge detecting operator. We propose a module called Sobel Heuristic Kernel(SHK). Our work makes several main contributions and experimentally shows good performance. To the best of our knowledge, we are the first to combine traditional edge detection method and deep learning method in semantic segmentation. Our SHK module reaches state of the art in the Inria Aerial Image Labeling dataset.},
added-at = {2024-02-22T15:24:38.000+0100},
author = {{Hu}, T. and {Wang}, Y. and {Chen}, Y. and {Lu}, P. and {Wang}, H. and {Wang}, G.},
biburl = {https://puma.ub.uni-stuttgart.de/bibtex/2855f46e3eda9f6752c6744e422001ec1/yaowang},
booktitle = {2018 25th IEEE International Conference on Image Processing (ICIP)},
doi = {10.1109/ICIP.2018.8451170},
interhash = {e0fb3a16d688ae07a02f7eaa2f42f521},
intrahash = {855f46e3eda9f6752c6744e422001ec1},
issn = {2381-8549},
keywords = {(artificial Aerial Detection Heuristic Image Kernel;Kernel;Semantics;Image Labeling Segmentation;Edge contour;aerial dataset;Aerial detecting detection detection;Neural detection;image edge image intelligence);Inria method;SHK module;Sobel networks;Detectors;Convolution;Semantic operator;traditional segmentation;Image segmentation;classical segmentation;learning segmentation;semantic semantic},
month = Oct,
pages = {3074-3078},
timestamp = {2024-02-22T15:24:38.000+0100},
title = {Sobel Heuristic Kernel for Aerial Semantic Segmentation},
year = 2018
}