频道关注基于GAN的合成杂草生成,用于精确的杂草识别
Tang Li1, Motoaki Asai2, Yoichiro Kato1
1Graduate School of Agricultural and Life Sciences, The University of Tokyo, Tokyo 188-0002, Japan.
Plant phenomics (Washington, D.C.)
|April 1, 2024
概括
这项研究引入了一种新的生成对抗网络 (CA-GAN),以创建现实的合成杂草数据. 这种方法有助于为数字农业开发特定地点的杂草管理 (SSWM),减少了对大量手动数据注释的需求.
科学领域:
- 农业科学 农业科学
- 计算机视觉 计算机视觉
- 人工智能的人工智能
背景情况:
- 杂草感染显著降低了作物产量,引发了人们对除草剂等传统杂草控制方法对环境的影响的担忧.
- 现场特定杂草管理 (SSWM) 对可持续农业至关重要,需要通过深度学习准确识别作物和杂草.
- 对于SSWM的深度学习模型,需要大量的,专业注释的数据集,这在开发中构成了重大瓶.
研究的目的:
- 开发一个能够产生高质量的合成杂草数据的生成对抗网络 (GAN).
- 为应对在农业应用中训练深度学习模型的有限注释数据的挑战.
- 加强对特定地点的杂草管理 (SSWM) 策略的开发.
主要方法:
- 针对合成杂草的数据生成,提出了一个通道注意力机制驱动的生成对抗网络 (CA-GAN).
- 该CA-GAN模型在两个不同的数据集上进行了训练和评估:分段植物播种数据集 (sPSD) 和可持续农业生态系统服务研究所 (ISAS) 数据集.
- 使用识别精度和Fréchet初始距离 (FID) 评分来评估性能,以衡量数据质量和与现实数据的相似性.
主要成果:
- 由CA-GAN生成的合成杂草数据实现了高识别准确度:sPSD上的82.63%和ISAS上的93.46%.
- 20.95 (sPSD) 和24.31 (ISAS) 的低频率发射距离 (FID) 分数表明合成和真实数据集之间存在很高的相似性.
- 拟议的CA-GAN在图像质量,多样性和可辨别性方面表现优于现有的GAN模型.
结论:
- 该CA-GAN模型有效地生成现实的合成杂草数据,克服了手动注释的局限性.
- 这种方法对推进数字农业和特定场地杂草管理 (SSWM) 系统有很大的前景.
- 该方法提供了一个可行的解决方案,用于创建大型,多样化的数据集,这对于培养农业中强大的深度学习模型至关重要.
更多相关视频
11:14Rapid High-throughput Species Identification of Botanical Material Using Direct Analysis in Real Time High Resolution Mass Spectrometry
Published on: October 2, 2016
11.6K
08:18Non-Aqueous Isolation and Enrichment of Glandular Capitate Stalked and Sessile Trichomes from Cannabis sativa
Published on: May 12, 2023
2.0K
相关概念视频
02:55Synthetic Biology
4.7K
Synthetic biology is an interdisciplinary science that involves using principles from disciplines such as engineering, molecular biology, cell biology, and systems biology. It involves remodeling existing organisms from nature or constructing completely new synthetic organisms for applications such as protein or enzyme production, bioremediation, value-added macromolecule production, and the addition of desirable traits to crops, to name a few.
Golden rice
Golden rice is a genetically modified...
Golden rice
Golden rice is a genetically modified...
4.7K
02:16Light Acquisition
8.5K
In order to produce glucose, plants need to capture sufficient light energy. Many modern plants have evolved leaves specialized for light acquisition. Leaves can be only millimeters in width or tens of meters wide, depending on the environment. Due to competition for sunlight, evolution has driven the evolution of increasingly larger leaves and taller plants, to avoid shading by their neighbors with contaminant elaboration of root architecture and mechanisms to transport water and nutrients.
8.5K