OStream, the object data solution provider for deploying efficient physical world AI, announced its software and data transport system, Percept™, to cut costs and slash the time-to-market for building rich media AI solutions using OStream enabled hardware and data management services.
Affordable, wirefree cameras and sensors are already widely available but their data load burdens customer networks, linearly drives up cloud costs and limits practical deployments. Percept integrates any camera with any AI service.
OStream Percept consists of:
- Virtual Object Camera that can be used as embedded software inside packaged, wirefree object cameras with 1km transmission range, connected to generic IoT cameras, and provides a software intake point for legacy wall-powered IP cameras;
- AI Hub server software with an AI Objects Router and Data Lake to integrate open AI services and provide storage and search;
- AI Services Market aggregating access to over 300 AI model and processing service providers plus webhooks for 3rd party platforms
The push for rich media AI processing at the edge raises equipment and maintenance costs while limiting sensor deployments to locations that have wall power and existing WiFi network reach. The OStream low power Object Codec converts rich media streams to compact objects, allowing AI candidate data to be captured and selected cheaply on battery powered edge devices, then sent over a low power, long distance wireless network.
The wirefree design allows Percept enabled devices to be deployed virtually anywhere and bypass customer WiFi, completely avoiding any server class computing on edge devices. In the cloud or on-premise, OStream deploys its AI Hub to aggregate candidate data from the IoT edge devices, route it to the solution builder’s chosen cloud AI service and store, retain and search data at scale in real time. The OStream Object Codec, low power wireless networking and AI Hub enables computer vision virtually anywhere, without relying on customer IT infrastructure.
Tiny Edge Computing Reimagines Allocation of Computing to Balance Speed, Cost & AI Power
The OStream Object Codec applies Tiny Edge Computing to edge cameras and other rich media instruments, where object conversion and lightweight AI candidate data is selected, slashing the necessary video and image data load by up to 97.5%. Low-power, wirefree networks can easily handle the reduced data traffic for sophisticated AI. OStream Percept hardware and software offers:
- Wirefree Computer Vision deployed with 25X the range of WiFi
- Object stream analytics previously unavailable or not feasible with streaming video
- Wirefree Computer Vision and rich media AI anywhere
- Lowest running costs of any rich media AI solution
- Pair any OStream device with any AI service
Rides on Ripl™
Once rich media streams are converted into compact objects and screened for candidate data to be pushed to the cloud, OStream incorporates Ripl™ Networks MLMesh™ for secure IP data transport over its dynamically-optimized wirefree network. Endpoints enjoy up to 1km of wireless network range that doesn’t compete with WiFi and bluetooth. End to end, your rich media data streams travel in IP IoT.
Open, Flexible OStream AI Hub
OStream, with its AI Hub, can support AI-in-the-cloud, on-premise, offline or in hybrid deployments. Wherever the Hub is installed and run, it directs candidate data to any preferred ML service regardless of location. The Hub can also manage instructions for candidate data storage in any destination, enforcing storage policies on a continuing basis. The Percept AI Hub is the initial destination for all Percept candidate data streams, before data is directed to AI services and storage destinations.
CEO Kerry Shih noted, “OStream removes the friction from deploying and operating a computer vision system. With wirefree device deployments and software that dramatically cuts down on data load, we enable customers to invest in infrastructure that puts them in control of where their object data gets routed to, digested and presented. It’s the simplest way to securely pair any rich media device with any AI service.”
OStream Devices Help System Integrators and Solution Builders Apply Rich Media AI to Everyday Business and Consumer Problems
Analysts agree that for Vision AI and related markets to reach their potential, applications must solve lasting deployment and security problems and be economically viable for routine business functions. Enabling inexpensive IoT edge cameras, LIDAR, microphones, etc. with low power wirefree networking and easy access to any preferred ML/AI processor allows integrators to solve vertical-specific problems. Retail, construction, crowd management, public safety, security, and hundreds of other solutions can be quickly developed and installed where wired networks, wall-power requirements and full rich media transport would otherwise be too difficult and expensive to deploy.
Peter Chambers, CEO of DTech (https://www.deployedtech.com/) has evaluated Percept. “We’ve seen vision AI be a powerful tool to secure and optimize hospitality and OStream’s Percept is the first solution that gives us the freedom to use any camera while leveraging a huge marketplace of AI Services,” said Mr. Chambers.
OStream Cameras and Other Media Devices Make Rich Media AI Instrumentation Easy and Inexpensive.
Until now, instrumentation complexities have inhibited computer vision deployments. OStream sells affordable edge hardware and long-distance access points under their own brand, but mainstream hardware providers will also be licensed to build their own Percept enabled devices. All such devices will send their AI candidate data streams to The OStream AI Hub, unless external providers license the Hub to be hosted elsewhere under their own brands. Watch for partner announcements pending.
The first “Percept Inside” cameras and associated devices will be available on March 21st from Mouser, with additional channels coming.
