Overlap Training to Mitigate Inconsistencies Caused by Image Tiling in CNNs
Files in This Item:
|Request a copy||Overlap Training to Mitigate Inconsistencies Caused by Image Tiling in CNNs_.pdf||1.08 MB||Adobe PDF|
|Title:||Overlap Training to Mitigate Inconsistencies Caused by Image Tiling in CNNs||Authors:||An, Yu; Ye, Qing; Guo, Jiulin; Dong, Ruihai||Permanent link:||http://hdl.handle.net/10197/12583||Date:||8-Dec-2020||Online since:||2021-11-03T09:07:46Z||Abstract:||This paper focuses on the problem of inconsistent predictions of modern convolutional neural networks (CNN) at patch (i.e. sub-image) boundaries. Limited by the graphics processing unit (GPU) resources, image tiling and stitching countermeasure have been applied for most megapixel images, that is, cutting images into overlapping tiles as CNN input, and then stitching CNN outputs together. However, we found that stitched (i.e. recovered) predictions have discontinuous grid-like noise. We propose a simple yet efficient overlap training framework to mitigate the inconsistent prediction at patch boundaries without changing the model architecture while improving the stability, robustness of the model. We have applied our solution to various CNNs (such as U-Net, DeepLab, RCF) and tested them on two real-world datasets. Extensive experiments suggest that the new framework is sufficient in reducing inconsistency and outperform these countermeasures. The source code and coloured figures are made publicly available online at: https://github.com/anyuzoey/Overlap-Training.git.||Funding Details:||Science Foundation Ireland||Funding Details:||Insight Research Centre||Type of material:||Conference Publication||Publisher:||Springer||Series/Report no.:||Lecture Notes in Computer Science; 12498; Lecture Notes in Artificial Intelligence; 12498||Copyright (published version):||2020 Springer||Keywords:||Convolutional neural networks; Computer vision; Image segmentation; Fault recognition||DOI:||10.1007/978-3-030-63799-6_3||Other versions:||http://bcs-sgai.org/ai2020/||Language:||en||Status of Item:||Peer reviewed||Is part of:||Bramer, M., Ellis, R. (eds.). Artificial Intelligence XXXVII: 40th SGAI International Conference on Artificial Intelligence, AI 2020, Cambridge, UK, December 15–17, 2020, Proceedings. SGAI 2020||Conference Details:||The 40th SGAI International Conference on Artificial Intelligence (AI-2020), Cambridge, United Kingdom (held online due to Coronavirus outbreak), 15–17 December 2020||ISBN:||978-3-030-63798-9||This item is made available under a Creative Commons License:||https://creativecommons.org/licenses/by-nc-nd/3.0/ie/|
|Appears in Collections:||Insight Research Collection|
Show full item record
If you are a publisher or author and have copyright concerns for any item, please email email@example.com and the item will be withdrawn immediately. The author or person responsible for depositing the article will be contacted within one business day.