I'm reviving these issues. Basically, we need example code on how to load in-memory images into pipeline (for ONNX and TensorFlow models):

#2495

#2121

There are many scenarios that require that way instead of loading from image files on the drive.

Scenarios:

• Live-streaming images coming from a laptop camera for object detection ONNX models
• Web/Service Http apps where images come through HTTP and need to be processed directly

Can the team create example code of ONNX model and a second for a TensorFlow model where in-memory images are provided instead of loading from files on the drive?