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.

煙葉分選

利用光譜成像看得寬,物質特征波段抓得準的特性,採(cǎi)集煙葉形狀、可見色澤之外的緻密度、含油含水率等關鍵特性,結合模型匹配算法,輸出分級分選識别信号,再結合工業自動化控制設備(bèi)或系統,實現精準、高效的煙葉在線分選。


系統示意圖


1.png 

成果展現


2.png


煙葉與塑料分選

高光譜成像在工業行業中具有多種應用,包括對生産線物品的品控、質量控制和對沒有視覺差異但具有不同化學成分的物質(例如塑料)進行識别、分類和篩選。

圖片3.png


煙葉與塑料有獨特的光譜特征,經過圖像處理/光譜分析/機器學習和訓練,可快速對不同物質進行分選

大米的快速分類

同形同色不同質的物品,通過光譜進行識别、分類和标識,有效區分“魚龍混雜”,相較傳統方式的小樣本抽檢,可以做批量檢測,在時間和準確率上有突出優勢。

圖片2.png


紅色:東北大米,綠色:江西大米

枸杞原産地溯源

與中國藥科院合作,通過高光譜成像系統,採集不同産地枸杞光譜特征數據,建立特征數據庫,利用高光譜成像系統分析軟件自動分析、學習、分類,實現對枸杞等中藥材原産地的追溯。同時,也方便各級用戶對藥材品質特征進行快速識别。

圖片1.png