At its core, MarshallAI is a deep learning company that helps users replicate their own sensing. Applications include machine vision with excellent object detection capabilities, automatic audio categorisation and radio frequency analysis. The technology is used in countless scenarios and provides the greatest return when being a seamless part of a broader business process.
This is why we focus on building comprehensive integrations to existing systems in addition to other forms of collaboration with partners. We work both with end customers and partners, and have no problem with white-labeling our solutions.
If you represent an integrator, a camera vendor, a telecom operator, an infrastructure provider, an IoT-device manufacturer, a private security company, or any other relevant domain and are looking for network architecture agnostic deep learning solutions, do not hesitate to contact us.
The MarshallAI AFSA (Agile Focused Signal Analytics) Platform enables users with fundamental computer skills to replicate their own audio-visual perception into an automated process on their own, without writing a single line of code. The platform runs on any public cloud, in your private data center, on the edge, in a hybrid environment, virtualized, or on bare metal.
The platform supports any visual use case including e.g. rtsp or rtmp streams, analog cameras or USB devices. In addition, the platform directly supports automatic analysis of audio or radio frequency spectrums.
Everything can either be controlled via our browser-based user interface or when needed, headlessly. The platform consists of four main sections:
- AI Models - the human eye or ear
- AI Programs – the way we act
- AI Processes – running AI programs
- Analysis – search engine for meta data
Create new AI models by using data-driven programming by dragging and dropping, without writing a single line of code. We have created a variety of preconfigured templates so you can quickly understand how it works before customizing your own project template according to your specific needs.
You can create new AI programs based on these templates and then customize them by adding configurable nodes via the drag & drop functionality into your own projects. This gives you full control over every aspect of your systems as well as all generated metadata. Our autotune mechanism makes the improvement of neural networks easy – we save you both valuable time and costs, and enable very fast bootstrapping of a new AI Model.
Our platform integrates seamlessly off-the-shelf with a number of third-party systems, to deliver relevant metadata in real-time. Integration nodes to new systems can easily be created as needed, even by yourself when preferred. A range of analytic tools is available to help you improve efficiency and accuracy in a semi-supervised manner. The platform is suited for:
- Visual information
- Audio inputs
- Radio frequency data
- Other unstructured data
Superior real-time object detection accuracy in any environment. State-of-the-art proprietary backend tools enable fine tuned models and tailored customer solutions by means of mere configuration.
Our platform works with any camera. No need for replacement or changes to the existing hardware. We only need access to the video feeds - the service can be up and running in days on any infrastructure, from private cloud to the edge.
We put the high value on people's privacy. We don't store footage or identifiable parts of data. Faces and similar elements are always anonymised. We require high ethics from our customers as well and never provide our technology to be used for any illegal activities.