
# EasyDeepLearningInferenceLive

This package contains a collection of samples using the Open eVision library. The goal of these samples is to show how
to use Open eVision in a live application on embedded platforms like smart cameras to do inspection with a Deep Learning
tool.

## Supported APIs

Embedded platforms and smart cameras often use specific programming interfaces to give access to images coming from
a camera. You will find samples in this package for the following programming interfaces:

* LibCamera ( https://libcamera.org/ )
* V4L2 ( https://en.wikipedia.org/wiki/Video4Linux )
* Pylon 5 ( https://www.baslerweb.com/en/products/basler-pylon-camera-software-suite/ )
* Neoapi ( https://www.baumer.com/be/en/product-overview/industrial-cameras-image-processing/software/baumer-neoapi/c/42528 )
* A fake camera using a single image
* A fake camera using a video read using the FFmpeg libraries (libavformat, libavcodec, libavutil, libswscale)

## Supported Displays

Depending on your use case you may need to have your application display your images or perform without a graphical
interface. To address those different use cases we provide each of our samples in the following versions:

* A Qt ( https://www.qt.io/ ) version displaying the images as well as the inspection result in a graphical window.
* A console version displaying only a summary of the inspection result.

## Supported Deep Learning tool types

All the Deep Learning tools are supported, i.e.:

* EasyClassify
* EasyLocate (Bounding Box and Interest Point)
* EasySegment Supervised
* EasySegment Unsupervised

## Which API to use?

The API to use is dependent on the type of system and the type of camera you intend to use. All embedded systems and
cameras may have different support.

Here is a table recapitulating systems and cameras that we tested and which API was confirmed to work on that system:

| System / Camera                                         | API known to work                      |
|---------------------------------------------------------|----------------------------------------|
| Raspberry Pi with MIPI CSI-2 camera                     | LibCamera or fake camera APIs          |
| Nvidia Jetson Developer kit with MIPI CSI-2 camera      | V4L2 or fake camera APIs               |
| Smart camera ADLink NEON-2000-JNX                       | Pylon 5 or fake camera APIs            |
| Smart camera Baumer Vax                                 | Neoapi or fake camera APIs             |


In this package, we provide two videos to use with the FakeCameraFFmpeg camera API:
* CeramicCapacitor.mp4 to use with the model found in `EasyLocate/CeramicCapa/Tool 1/CeramicCapa.edltool` in the Deep Learning
  Additional Resources.
* MiniWaffles.mp4 to use with the model found in `EasyClassify/MiniWaffles/Tool 1/MiniWaffles.edltool` in the Deep Learning
  Additional Resources.

## Copyright

All the document and source codes are the property of EURESYS S.A.
Copyright EURESYS S.A. 2026

The resources are provided “as is”, without warranty of any kind. In no event shall the
authors or copyright holders be liable for any claim, damages or other liability.
