基于WGAN-GP的建筑垃圾数据集的优化与扩充
Optimization and Expansion of Construction Waste Dataset Based on WGAN-GP
摘要: 随着人工智能技术的火爆与不断成熟，人们越来越倾向于用神经网络的方式去解决现有的问题。同时随着城市化的推进，建筑的拆毁与重建过程中产生了大量的建筑垃圾。现有的建筑垃圾回收装置回收工序复杂，效率低下，导致回收成本远远高于资源本身价值，因此探索高效率，低成本的建筑垃圾回收设备尤为重要。然而现在几乎不存在针对建筑垃圾的专有数据集，且大多数数据质量都不高，导致神经网络无法应用于建筑垃圾领域。为了解决上述问题，提高神经网络在建筑垃圾方面的应用，本文提出了一个新的建筑垃圾数据集，通过手工采集与清理的方式形成初始数据集，在通过WGAN-GP模型对其完成高质量的样本扩充，从而填补建筑垃圾数据集的空白。
Abstract: With the explosion and continuous maturity of artificial intelligence technology, people are increasingly inclined to use neural networks to solve existing problems. Meanwhile, with the advancement of urbanization, the process of building demolition and reconstruction has produced a large amount of construction waste. The existing construction waste recycling devices have complicated recycling processes and low efficiency, resulting in recycling costs much higher than the value of the resources themselves, so it is important to explore high-efficiency and low-cost construction waste recycling devices. However, there are almost no proprietary data sets for construction waste, and most of the data are of low quality, which makes it impossible to apply neural networks to the construction waste field. In order to solve the above problems and improve the application of neural networks in construction waste, this paper proposes a new construction waste dataset, the initial dataset is formed by manual collection and cleaning, and then it is expanded with high quality samples by WGAN-GP model to fill the gap of construction waste dataset.
文章引用：邬欣诺. 基于WGAN-GP的建筑垃圾数据集的优化与扩充[J]. 计算机科学与应用, 2023, 13(1): 136-142. https://doi.org/10.12677/CSA.2023.131014

参考文献

[1] 韩九强, 胡怀中, 等. 机器视觉技术及其应用[M]. 北京: 高等教育出版社, 2009.
[2] 杨凯. 建筑垃圾的现状分析及资源化广泛利用必要性[J]. 建筑工程技术与设计, 2017(21): 4544.
[3] 李景茹, 朱姣兰, 蔡红, 等. 设计人员的建筑废弃物减量化行为影响因素分析[J]. 科技管理研究, 2014, 34(14): 183-187.
[4] 高天霞, 王海兵. 工业化建筑及其废弃物减量化探讨[J]. 广州建筑, 2018, 46(1): 35-40.
[5] 陈家珑. 我国建筑垃圾资源化利用现状与建议[J]. 建设科技, 2014(1): 8-12.
[6] 牛斌, 吴鹏, 马利, 等. 一种基于生成对抗网络的行为样本集扩展方法[J]. 计算机技术与发展, 2019(7): 43-48.
[7] Denton, E.L., Chintala, S., Szlam, A. and Fergus, R. (2015) Deep Generative Image Models Using a Laplacian Pyramid of Adversarial Networks.
[8] Goodfellow, I.J., Pouget-Abadie, J., Mirza, M., et al. (2014) Generative Adversarial Networks. Advances in Neural Information Processing Systems, 3, 2672-2680.
[9] 段峰, 王耀南, 雷晓峰, 等. 机器视觉技术及其应用综述[J]. 自动化博览, 2002, 19(3): 59-61.
[10] 杨壮. 面向Bin Picking的虚拟样本集构建及智能识别方法的研究[D]: [硕士学位论文]. 上海: 华东理工大学, 2019.
[11] Arjovsky, M., Chintala, S. and Bottou, L. (2017) Wasserstein Generative Adversarial Networks. Proceed-ings of the 34th International Conference on Machine Learning, Vol. 70, 214-223.
[12] Gulrajani, I., Ahmed, F., Arjovsky, M., Dumoulin, V. and Courville, A. (2017) Improved Training of Wasserstein GANs. Advances in Neural Information Processing Systems 30: Annual Conference on Neural Information Processing Systems, Long Beach, 4-9 December 2017, 5767-5777.
[13] 章毓晋. 图像理解与计算机视觉[M]. 北京: 清华大学出版社, 2000.
[14] 王志瑞, 闫彩良. 图像特征提取方法的综述[J]. 吉首大学学报(自然科学版), 2011, 32(5): 43-47.
[15] Lowe, D.G. (2004) Distinctive Image Features from Scale Invariant Keypoints. International Journal of Computer Vision, 60, 91-110. [Google Scholar] [CrossRef]