The Rise and Rise of Computervision Part 2 – The one system approach
Audio Blog Version:
Introduction
In the first part of this blog, we talked about the relevance and growth of the use of computer vision technologies across various industries and how it is adopted to help in our everyday lives. Organisations across industries are increasingly looking at AI-driven technologies to analyse images and videos—often in real-time—to solve a variety of problems such as enhancing customer experience, increasing safety, and improving product quality to name a few.
When it comes to choosing the one you need it will always tend to be business need led but you also need to consider your strategic approach. For example, in Retail you may be looking at demographics, footfall or scan avoidance to name a few use cases but you need to consider what you may also need in the long term too, zone management, integration with signage, pallet counting in the warehouse etc.
Some companies will adopt a strategy of working with one provider in a specific industry. We are now starting to see companies who can fulfil this end-to-end strategic need by taking more of a one system approach allowing you to start small fulfilling specific use cases and then grow with you as your use cases widen across a whole organisation.
The one system approach
We have already talked about how commercially available models are the way forward. What if you could also import your existing models as well? What if the model is one that needs training significantly and you haven’t got enough data to train it accurately? What if the model you need, needs customisation?
Sounds daunting…but it does not need to be. Companies like meldCX have taken the one system approach with their platform Viana. Their approach allows you to scale the solution as potential use cases grow. It comes in various levels as you will see below.
Before going into the detail lets understand why the one system approach can be of advantage:
- Scaling your use of Computervision AI as your business grows as well as your use cases
- The advantage of one system for your team makes it easier to understand, learn and develop in the medium and long term with the added benefit of streamlining your costs
- Build out of the box all the way through to customised models
- Taking advantage of synthetic data modelling when you don’t have enough data to train a customised model
- Lower the barrier to entry for AI
- Ability to integrate your own models
- Less infrastructure disruption
- Choice of deployment, on the edge, on-premise, in the cloud or hybrid
The meldCX Approach
meldCX with their platform have built a variety of models which you can use and scale with.
Viana Lite is the perfect balance between performance and price. It comes with an easy-to-use software interface and can be used in almost any industry.
Viana Standard adds more advanced features like multi-camera support, object tracking and more powerful analysis capabilities for the ultimate in customisation.
Viana Premium comes loaded with everything you need for rapid deployment across your enterprise. This model has been specially designed to meet the needs of large corporations that have complex production lines or require high throughput speeds without compromising accuracy or uptime requirements.
Finally, Viana Enterprise offers customisable options unavailable elsewhere on the market today so that no matter what type of application or business problem you’re facing they have got a solution just right for you!
Walk Through of a typical approach with Viana
Viana lite is suited to those who want to kick off with a small number of models, or for those who have never used software before and want to test the water before diving in. Viana Lite is also good for testing out the platform and seeing how it works without having to commit too much time and effort.
Viana Standard offers a range of generic image recognition models to help you get started. It’s great for companies looking to get started with image recognition without having to build their own custom model from scratch.
Viana Premium – Take advantage of a range of packaged models, ready to deploy as they are or customised to your needs. Viana Premium is a fully integrated platform and enables you to build new models or combine existing ones, store and manage training data, and deploy edge devices globally. This solution offers you maximum flexibility in terms of what you want to do with it.
Viana Enterprise – Is a fully integrated platform enabling you to build new models or combine existing ones, store and manage training data, and deploy edge devices globally.
With Viana Enterprise you can:
- Build new models or combine existing ones into one scalable solution.
- Store and manage training data from your devices without needing to worry about network bandwidth limitations.
- Deploy devices across multiple geographies on-premises or in the cloud with managed security features already built-in by meldCX (no need for additional hardware).
Let’s now consider if you want to take your computer vision further. Companies like meldCX have two other levels Custom and Train that can give you the ultimate in how you need to scale your strategy.
Viana Custom – Build new visual models or integrate existing ones into one platform, creating your own end-to-end solution with cross-platform deployment.
- Build your own models
- Integrate existing models
- Create your own end-to-end solution
- Cross platform deployment
Viana Train – Use cutting-edge synthetic data modelling to create unlimited datasets that can be easily stored, managed and enhanced. The platform is built on a distributed architecture and provides flexibility in terms of storage, management and enhancement of your data. This allows you to increase the size of your original dataset by adding images from multiple sources. This feature is useful where you have an existing training set but need an additional test set or vice versa. You can also use this feature if you want more data for model building purposes or simply want to test out different algorithms on new datasets. Another reason why this feature is beneficial for machine learning projects is because it allows users to easily store their images when they’re not being used for training or testing purposes which reduces storage costs considerably compared with storing all images locally at once (e.g. on hard disk).
meldCX has a solution for a one system approach
meldCX has a one system approach solution for all your vision analytics requirements. They are an Intel partner, with a wide range of products and solutions to fit your needs. The meldCX platform provides one central location where you can deploy and manage all your computer vision applications across multiple industries.
Conclusion
Whether you want to start small or scale up as your needs grow a one system approach may be the answer to your vision analytics strategy.
Whichever approach you take, point solutions, industry solutions or one system solutions, the key is to work with solutions or organisations that will provide models with high levels of accuracy, is available within a commercial framework and most importantly have done it before and you can see how existing customers are using the technology.