[ Tech stack ]
OpenVINO
Intel's toolkit that accelerates inference on CPUs, integrated GPUs and NPUs.
OpenVINO converts a PyTorch, ONNX, TensorFlow or TFLite model into its intermediate representation (IR) and runs it through hardware plugins: Xeon and Core CPUs, integrated or Arc GPUs, the NPU in Core Ultra chips, with AUTO and HETERO modes to spread the load. Its NNCF compression framework quantizes to INT8 or compresses weights to INT4, OpenVINO Model Server exposes models over gRPC/REST, and OpenVINO GenAI serves LLMs locally. It's our choice when inference has to run on Intel PCs, industrial mini-PCs or servers without a dedicated GPU.
[ Why OpenVINO at Dexon ]
What this technology does well, and why we use it.
Typical usage: Industrial vision on mini-PCs, smart checkout or kiosks, local LLM assistants on workstations.
- 01
One plugin per Intel target: CPU, integrated GPU, Arc or the Core Ultra NPU.
- 02
Direct conversion from PyTorch, ONNX, TensorFlow and TFLite to the IR format.
- 03
NNCF: INT8 quantization and INT4 weight compression, LLMs included.
- 04
Open source (Apache 2.0), gRPC/REST Model Server and OpenVINO GenAI for local LLMs.
[ Complementary technologies ]
The building blocks we often mobilise alongside.
A stack rarely exists alone. Here are the technologies Dexon most often pairs with this one, through pipeline habits, usage similarity or internal mastery. Click on a building block to see its scope.
[ Reassurance ]
- 0+
- custom projects delivered
- 30
- engineers, designers, project managers
- 80 %
- from top French schools
- 24 h
- average reply time
[ They trust us ]
More than 100 French and European companies trust us
[ Press ]
They talk about us.
Nationwide coverage by BFM Business, Le Figaro, Challenges, La Tribune and CNews. An outside reading of our work and our innovations.
