OSG(a)i: because AI needs a runtime
summary
Claim: OSGi functions as a modular runtime for trained models by packaging a PyTorch-trained MNIST network as an ONNX export inside a bundle with metadata (input shape, output classes, version), exposing it as an OSGi service and resolving it at runtime. The demo converts Python model → ONNX → jar, uses DS and PushStreams processing to map camera frames → tensor → model → arm controller, and shows a robot reliably picking a blue block using promises and remote OSGi services.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 1:10What is AI
- 1:33Machine Learning
- 3:37Frameworks
- 4:42Data Scientists
- 7:17TensorFlow
- 8:57Check the boxes
- 11:50Read the dump file
- 12:48Prototype framework
- 13:50Demo
- 19:11Use case
- 20:25Integration with neural nets
- 24:08Conclusion
related talks
05mONNX Runtime
51mMaking neural networks run in browser with ONNX
30mThe Evolution of ONNX: Driving Interoperability in AI...
30mScalable ML acceleration with ONNX Runtime
33mWhat on earth is ONNX and ONNXRUNTIME
45mLeverage Power of Machine Learning with ONNX
05mONNX-Go, neural networks made easy
05mPyTorch ONNX Export Support