As a developer or system integrator of AI video applications, are you concerned that you don’t have enough tools and resources to build a powerful AI solution? Are you struggling with the uncertainty of the return on investment (ROI) of your edge AI projects? As a developer of computer vision or AI models, do you face the challenge where specific algorithms can be implemented easily, but building a full system or product around them is difficult? As a hardware manufacturer of AI boxes or AI servers, are you worried about slow product deployment because there is no end-to-end software solution to integrate your product?

Users and developers can download BrainFrame onto your Linux machine from Aotu’s website at https://aotu.ai/docs/getting_started/. No code is needed. You can use prepackaged AI algorithms from the VisionCapsule AI Marketplace with a click, or follow the OpenVisionCapsules format—co-released with OpenCV.org—to deploy your own AI algorithms. You can integrate BrainFrame with your own applications through REST APIs.

Input video, output structured data. Whether using edge computing or cloud deployment, plug-and-play prepackaged AI algorithms from the Marketplace are ready for use in safety and security applications and various other use cases.

• BrainFrame runs video processing and AI inference on either edge or cloud computing devices, extracting real-time insights from IP cameras or any video source for continuous monitoring and analysis tasks such as detecting, recognizing, classifying, tracking, and counting people, vehicles, objects, text, etc.
• BrainFrame produces structured data output in the form of graphical reports and real-time alerts, allowing developers and system integrators to process multiple tasks for broad inspection and monitoring use cases. In addition, it provides all the tools necessary for large-scale deployment, such as:
  • Database management
  • System tools to integrate with cameras and network video recorders
  • REST APIs for system integration or to build any AI application on top of it

• BrainFrame comes with the VisionCapsule AI Marketplace, which includes many prepackaged AI algorithms for out-of-the-box smart vision applications*, such as:
  • Administration: Queue/occupancy, shelf space/inventory management
  • Industrial: Production line defect detection, safety compliance monitoring
  • Retail: Retail/restaurant, bank, and store management
  • Traffic/Urban: Management and analysis
• Users can configure regions, lines, and alarms with AI and video overlays based on their use case needs. The configuration and graphical reports can be adjusted on the go to fit specific requirements: https://aotu.ai/docs/dashboard/getting_started/

NO CODE. Fast to deploy. Easy to use. Ready to scale.

• BrainFrame’s What You See Is What You Get (WYSIWYG) Graphic User Interface is designed for non-engineers. Simply install BrainFrame—it’s ready for use with no programming required. With the recommended computer hardware (https://aotu.ai/docs/user_guide/server_hardware/), software installation on x86 computers takes less than one hour. Just enter the IP address of the IP cameras, configure the areas of interest with smart lines and zones, and input your email address or phone number to start receiving alerts and reports. The system configuration for cameras usually takes only a few minutes: https://aotu.ai/docs/getting_started/#install-brainframe-client
• If the computer has GPUs installed, any high-performance, high-accuracy VisionCapsules can be used. If there is no GPU, OpenVINO-compatible VisionCapsules can be used.
• BrainFrame can operate as a standalone AI application for a few camera streams, or as an enterprise/industrial video analytics service with clustering and multi-premise* support for thousands of cameras.

LOW CODE. Easy to customize and integrate.
BrainFrame includes many vision applications that users can install and immediately begin using—no programming required. It also has powerful customization support to extend its capabilities to meet unique needs.

• Developers can integrate BrainFrame into a system or build applications based on BrainFrame’s REST API and database API: https://aotu.ai/docs/api/
• BrainFrame’s VisionCapsules utilize the OpenVisionCapsules algorithm packaging format. Developers can build their own VisionCapsules using this format and deploy them with BrainFrame. The VisionCapsule runtime engine can automatically fuse these custom Capsules with prepackaged ones from the Marketplace, allowing developers to focus on the unique models or algorithms for their use case while leveraging existing Capsules. It usually takes fewer than 50 lines of code to develop a new end-to-end VisionCapsule by referencing the sample code. OpenVisionCapsules is an open-source, BSD-licensed system for encapsulating machine learning and computer vision algorithms, co-released by Aotu and OpenCV. The source code is available on OpenCV’s GitHub: https://github.com/opencv/open_vision_capsules

BrainFrame features advanced AI technologies. Its powerful VisionCapsule inference engine supports automatic algorithm fusion, parallel inference scheduling, and computing across CPU, GPU, iGPU, NPU, etc.

• BrainFrame’s AI inference pipeline and scheduler are fully optimized for video streaming on x86 platforms and NVIDIA GPU architectures.
• BrainFrame is compatible with Intel CPUs, Intel Graphics, Intel NPUs, and NVIDIA GPU AI acceleration platforms. It can also support other silicon vendors’ solutions via its VisionCapsule runtime architecture.
• Users can load or unload VisionCapsules as needed. Capsule loading and data/algorithm fusion are fully automated and happen at runtime. Loading/unloading does not interrupt ongoing operations and meets telecom operation standards.
• If a new vision algorithm is needed for a unique use case, developers can reuse prepackaged AI algorithms from the Marketplace and focus development on the unique algorithm. The VisionCapsule runtime engine automatically integrates the new algorithm with existing Capsules, dynamically forming an inference pipeline. For example, if developing a new object detector, reuse an existing tracker from the Marketplace.
• A fully functional GUI lets users control VisionCapsule deployment—select which streams a Capsule is activated for and control runtime parameters.

For Hardware OEMs and Enterprise Data Centers:

• OEMs can preinstall BrainFrame software on edge AI computing devices.
• Enterprises can deploy BrainFrame in their data centers to provide a centralized video and image AI platform.

BrainFrame Downloads and Documentation:

• https://aotu.ai/docs/downloads/
• https://openvisioncapsules.readthedocs.io/en/latest/