Make dynamic vision AI that works

Robovision offers a vision AI platform that covers the full AI life cycle. Streamline the entire process of developing, implementing and adapting AI in today’s ever-changing business environment. 

Talk to an expert

Unlock new intelligence from visual data

Robovision is an intuitive platform that allows users, without a background in engineering or artificial intelligence, to develop and maintain AI-based vision applications autonomously. Unlock new intelligence from visual data, as Robovision facilitates the entire lifecycle, from development to implementation to scaling out. With minimal iteration times: update, improve or switch between different models if there are new data or product types. Your business can quickly adapt to changing conditions or expand AI applications without technical hurdles.

AI to production with a single platform

By removing the downtime risk and costly change orders caused by dependency on IT or external consultancy services, our customers are continuously discovering how to leverage vision AI and the Robovision Platform for more cost savings and revenue growth. 

Dynamic AI that adjusts as conditions shift

With an advanced yet easy-to-use toolset, teams can take full control of their AI initiatives in a collaborative working environment for both domain experts and data scientists. AI applications that are theoretically rigorous, can be effectively developed and adapted to respond to continuously evolving business needs. 

Ensure business continuity & successful AI in production

Operationalising AI represents a high level of complexity. Robovision enables companies to move large-scale AI into production successfully. Teams can maintain a large number of applications, as well as adapt, improve performance and maximise their AI potential, all with minimal engineering support. 

Shorten AI Development Time

No matter where companies are in their AI adoption journey, Robovision helps speed up vision AI development. In an organised interface, accelerate the annotation and training process with powerful labelling capabilities, built-in algorithms and industry-standard evaluation metrics. Then finish the project by giving end-users the ability to adapt models in just a few clicks.

Bridge the Last Mile of AI Delivery

For  latency-critical operations, we have developed Robovision Edge, a platform add-on that delivers real-time insights for on-premises AI solutions. The Edge device will communicate with the Robovision Platform when models need to be deployed and adapted. 

Build your own IP Library

High-end industries face highly specific and complex AI challenges, often associated with the knowhow and IP of competitive advantage. Our platform supports a large number of built-in algorithms for a wide range of applications. Robovision is also easily extensible with specific algorithms through the Robovision Algorithms SDK.

Powering vision applications with APIs

Robovision API enables companies to create integration with other software and third party APIs quickly and easily. The API can be used to import and export datasets, control machines connected with edge devices, display results on a dashboard or connect with a robotic brain to execute tasks. The possibilities are endless!

How Robovision Platform works

A project is a workflow that manages the creation and training of the AI model. Companies can create projects of different types. The project types are grouped by a computer vision problem they tackle with machine learning algorithms, ranging from standard to customised, depending on the aim of your model; or in other words, how companies want the machine to see. 

Our standard projects include:

  • Image classification
  • Object detection
  • Instance segmentation
  • Semantic segmentation
  • Object tracking

The next step is to collect and upload samples, which will form the input for training the model. For example, if companies want to train a model to identify a defect in an image, it is important to have many images containing defects. The Robovision Platform enables data upload from local computer storage, cloud storage or Robovision Edge.

With our easy-to-use 2D and 3D labelling tools, domain experts can work with complex samples and apply their knowledge to complete the labelling.

For large and collaborative projects, assign labellers to share the workload and let them work closely together with experts to deliver AI. 

Once the basics are set up, train the model with the dataset containing labelled samples. The Robovision Platform allows multiple models to be trained simultaneously by different members of the team. This makes it easy for companies to manage and scale AI operationally. 

Robovision platform lets prolect leaders evaluate models easily from a single-intuitive interface

Let the Robovision platform work its magic and just like that, a trained model is developed. All is left is to check the accuracy of the model: Did the model annotate the images correctly? If there are any errors, then just pump more samples into the model and train it again. Project leaders or team leads can approve or reject models before deployment. 

Robovision offers 2 methods for deployment to meet different needs. 

Cloud deployment:

Via an API endpoint, the model is made available for use on your system, without complicated configurations.

 

Edge deployment:

Robovision Edge offers a quick and easy way to to connect trained models with on-site data for real-time inferences.

Robovision technology, software, AI solutions and algorithms are patent pending in various jurisdictions. Read more on our patents page.  

Key features

No-code technology

Perform every step of the AI life cycle, without ever encountering a line of code. Domain experts and business users can contribute their product knowledge and help develop better AI, without having to understand the ins and outs of it.

Advanced annotation tools

Annotate a wide range of visual data for basic to advanced annotation tasks with the platform’s intuitive yet powerful set of 2D, 3D and DICOM labelling tools. Rely on the intelligent grabcut and magnetic lasso to quickly annotate all sorts of irregular shapes.

Predictive labelling

Greatly speed up the annotation of large datasets with the platform’s predictive labelling feature: train a model on some already labelled data and use it to automatically predict the labels for the rest of the dataset.

Built-in algorithms

Build models with our out-of-the-box training algorithms that cover all common applications in production. Our anomaly detection algorithm can create a viable model with as little as 50 good training samples.

Operational independence

Companies can easily manage their project pipelines. Configure custom user roles and grant them access to different modules and functionality of the platform. Or create teams of users who work on a particular project. They can adapt only assigned models and access the required labelling toolset. 

Easy retraining

Most models work in dynamic environments where data is constantly changing. Retraining helps maintain the models’ accuracy and reliability. With Robovision, simply change the training data set and rerun the same process that generated production models. 

Integration-first

The final step in the end-to-end workflow is to bring AI to the real world by closing the deployment gap. Robovision offers extensive platform and Edge integration methods with both SDK and API integration possibilities. Cloud-based, on-premises or as part of a customer’s system, companies can easily implement AI models with our flexible deployment options. 

laminate-manufacturing-at-Unilin-Group

What you can develop with Robovision Platform

The Robovision platform allows companies to develop vision AI models with much more diverse goals than only solving simple problems. Here are some practical examples of how our platform has helped businesses across industries create highly personalised AI models that are beneficial to their companies in a large array of ways.

Helping market leaders revolutionise Agtech with vision AI

Staying ahead of competitors with intelligent systems, companies use Robovision to build the world’s first fully automated machine that recognises, picks and pots plant cuttings.

Driving automation in manufacturing processes

To automate hard-to-see defect detection, manufacturers streamline the visual inspection process with Robovision and vision AI to achieve a higher production output.

Assisting medical experts with AI-based diagnosis

To improve and speed up image-based diagnosis, doctors are assisted by AI models developed on the Robovision Platform that can identify specific organs and detect any anomalies.

Redefine your AI approach with Robovision

Start here