400-008-12207×24 service

Hyperspectral cloud platform for big data analysis and application


Apart from front-end equipment, powerful cloud computing platforms and algorithms are needed to promote the use of hyperspectral imaging. Cloud network platforms and services for hyperspectral image analysis utilize the parallel processing of multi-cluster CPU/GPU as well as the large storage and distributed computing capacity of clusters to boost processing efficiency and perform complex computations in a timely manner.

Using advanced technologies such as distributed storage, distributed computing and deep learning, Wayho's cloud platform for hyperspectral big data analysis provides users with data storage, data model training and comprehensive cloud-based solutions with hyperspectral technology for various industries.
Algorithms in
hyperspectral data
analysis services
One-stop
services
Resource sharing
services
Open source
spectral ecosystem

One-stop Services


Data storage, data pre-processing, data labeling, data model training and cloud deployment are integrated
to improve and shorten the product development cycle, as well as cut costs.

Spectral algorithnm engine service


The platform is built with our computation engine that incorporates big data analytics in hyperspectral imaging and a sharing mechanism to support high-throughput and high-efficiency calculations. It can also conduct real-time calculations of hyperspectral data with minimal delay, provide third-party access to algorithms, support combined data scheduling algorithm and building of complex spectral applications to provide developers and users with a solid foundation to process data and integrate applications.

Sharing resources to create
an open source spectral ecosystem


The platform provides the sharing and trading of spectral datasets, algorithms, models, solutions and other resources and services to promote the
rapid development of hyperspectral imaging, and create an intelligent open source spectral ecosystem.


Spectral datasets
Algorithms
Models
Solutions

Fexible and easy to use
with visual mode


The drag-and-drop configuration allows users to combine multiple algorithms and perform complex calculations.
It is also equipped with an easy-to-use self-adaptive system that displays the calculated results.

Safe and reliable


The platform uses a reliable security authentication mechanism with encryption to ensure user security and data confidentiality.

烟叶分选

利用光谱成像看得宽,物质特征波段抓得准的特性,采集烟叶形状、可见色泽之外的致密度、含油含水率等关键特性,结合模型匹配算法,输出分级分选识别信号,再结合工业自动化控制设备或系统,实现精准、高效的烟叶在线分选。


系统示意图


1.png 

成果展现


2.png


烟叶与塑料分选

高光谱成像在工业行业中具有多种应用,包括对生产线物品的品控、质量控制和对没有视觉差异但具有不同化学成分的物质(例如塑料)进行识别、分类和筛选。

图片3.png


烟叶与塑料有独特的光谱特征,经过图像处理/光谱分析/机器学习和训练,可快速对不同物质进行分选

大米的快速分类

同形同色不同质的物品,通过光谱进行识别、分类和标识,有效区分“鱼龙混杂”,相较传统方式的小样本抽检,可以做批量检测,在时间和准确率上有突出优势。

图片2.png


红色:东北大米,绿色:江西大米

枸杞原产地溯源

与中国药科院合作,通过高光谱成像系统,采集不同产地枸杞光谱特征数据,建立特征数据库,利用高光谱成像系统分析软件自动分析、学习、分类,实现对枸杞等中药材原产地的追溯。同时,也方便各级用户对药材品质特征进行快速识别。

图片1.png