Compare OpenVisionCapsules with TensorFlow, PyTorch, Keras, Caffe, ONNX, etc. formats.

OpenVisionCapsules is a portable NN and algorithm format for TensorFlow, PyTorch, Keras, Caffe, ONNX, etc.—all formats are supported. Through an encapsulation of the other formats and defining a standardized input, output, and configuration, OpenVisionCapsules is a self-contained format and interoperable smart-vision standard. It is designed for both developers and non-developers to use.

  • Enable mass production of AI
  • Drag-and-drop portable algorithms format to deploy
    • Self-contained; downloadable AI, good for both technical developers and non-technical users
    • Nearly-instantaneous load or upgrade at run-time
  • Tiny footprint
    • ~50 lines code: a few hours to learn

OpenVisionCapsules has been contributed to OpenCV.org, adopted as an industry standard for the purpose of:

  • Open standard for interoperability across machines, from edge to cloud
  • Agnostic to the underneath hardware architectures and machine learning frameworks
  • Negligible computing overhead and tiny footprint for embedded systems with limited computing & memory
  • Encryption to protect IP of algorithms and Neural network models
Compare neural network model formats with the OpenVisionCapsules format:

OpenVisionCapsules architecture

By encapsulating existing neural network models into OpenVisionCapsules format, you will be able to define the input/output type and pre/post-processing logic. In addition, you can configure the algorithm with a standardized API at runtime. With all the above information encapsulated, the OpenVisionCapsules format becomes completely self-contained. Thus it enables a non-technical person to deploy Neural networks or algorithms without any coding.

Resources

OpenCV.org/OpenVisionCapsules is released as a BSD license. The source code is available here:

https://github.com/opencv/open_vision_capsules

OpenVisionCapsules downloads: https://aotu.ai/docs/downloads/

Open-sourced OpenVisionCapsules: https://github.com/aotuai/capsule_zoo.git

OpenVisionCapsules documentation: https://openvisioncapsules.readthedocs.io/en/latest/

OpenVisionCapsules development tutorials: https://aotu.ai/docs/tutorials/capsules/installing_a_capsule/