Bee Yard



Machine vision framework and infrastructure for the whole image processing project lifecycle


Bee Yard



Machine vision framework and infrastructure for the whole image processing project lifecycle

Bee Yard



Machine vision framework and infrastructure for the whole image processing project lifecycle


Successful implementation of a machine vision project is a complex task. Everyone, who experienced the exciting journey knows how important are data & time.
The journey starts with the feasibility assessment. A phase that quickly reveals whether the project at hand has a reliable solution.
Following up is the development and implementation of robust algorithms that can cope with different environmental conditions.
Last but not least is the ability to pinpoint possible performance degradation ahead of time.
Our tools are fine-tuned to help you with every step of this complex process and enable you to gather high-quality data in a time-effective manner.

Typical machine vision workflow

Feasibility study

The crucial part of every project is a good assessment of scope at the beginning. For the vision projects, there are a lot of factors that are affecting the project feasibility such as imager, optics, lighting conditions, optical properties of the product and many more. It is a common practice to perform a feasibility study at the beginning of the project. The dream of every project manager and developer is to be able to get feasibility studies done in the shortest possible time. Our framework can help you with that.

Data collection
With our tools, you can easily store the data directly to the database.

Annotation
Offload data annotation from your developers and let them focus on the development. You can use our services or your own employees for annotation.

Algorithm development
Use our project templates to kick start your development. Our libraries provide an easy to use database interface to get the required data when you need them directly to your algorithm.

Collaboration
All participants can collaborate on a single database. One person can collect data, others annotate them, developers can start basic algorithm tests, sales guys can present data and the results to the customer and all this can be done in parallel!

Reporting
Keep your customers in the development loop by easy performance reports. Give customers direct access to their data and discuss the current project state.

Prototype

Every system is as strong as the weakest link. You certainly know the situation when your HW is producing great images in the factory and your developers are sitting in the office without any data to test on. Once you obtain the data you have to manually process them and go through piles of the data to locate correct data. Not anymore! Our infrastructure lets you setup local storage or direct connection to the cloud. With advanced data management your developers are always served with correct data that can be used for development.

Onsite data collection
Local or cloud infrastructure for data collection, management and transfer.

Automatic annotation
Based on machine or recipe you can annotate the data as they are collected and later on use this information in the tests.

Algorithm development
All your developers can collaborate on one ground-truth dataset in parallel.

Algorithm crossvalidation
The Bee Yard infrastructure provides complex tools for algorithm validation against the whole dataset. This task can be deployed to the cloud infrastructure.

Visualization
Hive database is providing customizable web viewer that can be used for result inspection.

Commissioning
Since the whole production is stored in the Hive database you can automatically generate performance reports and use them to validate your solution together with the customer.

Product

The process of data management and remote support starts to be much more challenging once your systems are deployed worldwide. Even in these situations Bee Yard comes to rescue.

Deployment
Support for remote production monitoring.

Production log
If your system is connected to the Hive database you can provide the customer with the full production log. You can mix raw data of failed inspections together with preview data of passed inspections.

Online annotation
Machine running the inspection, human operator or quality manager they all can label the data as they are being produced. Such rich information can be later used for algorithm improvements.

Algorithm customization
Thanks to Hive database you share the common interface with the customer. Customer can send you annotated data from the system and you can easily work on algorithm failures or new features.

Lifetime support

Once your product or project is up and running worldwide it is time to give the customer full support and stay ahead of your competition with new versions and features. With Bee Yard you can leverage data you are collecting from your systems. You can add data mining features and provide the customer with online statistics and reports as well as with updates for your product. With full history and algorithm cross-validation, you can be sure that you always deliver a working solution.

Data mining
Use prepared metrics or use prepared API to tailor your own data views.

Remote support
Reduce the cost of the system support by incorporating it into the product using the Bee Yard infrastructure.

Reliable updates
Cross-validate your algorithms on historical data to ensure stable production.

Advanced data structure

Industrial data structure evolved by years of experience. Our data "Cell" is optimized for mixture of data you would expect in every production line, such as images, binary data, calibration, text files, JSON files and many more. Moreover, you can label the cell with custom tags and even use vector overlays for the result visualization. These tags together with text files, JSON files and overlays can be used to search your data.

Advanced data structure

Bee Yard Infrastructure

Our infrastructure is covering the whole development scenario. You can start with data collection and data transfers to your cloud or server. Then you can parallelize your data annotation. Once you have the first annotated data your programmers can start immediately the development on the cell subset. Once the algorithms are working on the subset, programmers can deploy the algorithms to the swarm of workers and cross-validate their algorithm on the whole dataset. Developers can meanwhile work on the other tasks and analyze the results once they are ready.

 

Development process

Along the way, we found out that our developers are spending too much time organizing the data and waiting for the results of their algorithms. But not anymore! Thanks to the Bee Yard they can quickly select important data and focus on algorithm development. Once the algorithm is working on the selected subset developers can deploy the algorithms to the cloud. Meanwhile, their computer is ready for another tasks. At the moment the results are available, they are served via database API directly to the developer who can quickly identify the areas for the improvement. Once the algorithm is working on the entire data set, the developer is confident that it will work for the customer as well.

Development process

Developed by developers for developers!

Reach out
Developed by developers for developers!
Developed by developers for developers! Developed by developers for developers! Developed by developers for developers! Developed by developers for developers!

About Us

Company

Careers

Policies

Legal

Privacy

Contact

Follow Us

Email: info@datavision.cz
Phone: +420 212 340 365

DataVision s.r.o.
ID: 27142663
VAT ID: CZ27142663