v0.5.3

Release v0.5.3 includes:

• Fix linkage issues for the TFLite backend.

v0.5.2

Release v0.5.2 includes:

• Fix heisenbug where the tensorflow backend could segfault
  when configuring GPU memory usage if the default value was
  used.

v0.5.0

Introduced features:

• Add RAM usage property to tensorflow backend
• Check if the model is quantization is float32 on tflite

Known issues:

• NCSDK does not support multiple calls to the inference engine from the same thread. This causes the NCSDK backend to fail after the second start.

v0.4.2

Bug fixes

• Hardcoded size of output tensor

v0.4.1

Bug fixes

• Wrong number of predictions from Tensorflow Lite

v0.4.0

Introduced features:

• Multiple channel support
• GRAY8 support
• TFLite Backend support and examples

Known issues:

• NCSDK does not support multiple calls to the inference engine from the same thread. This causes the NCSDK backend to fail after the second start.

v0.3.0

Introduced features:

• Examples updated:
  • Inception for NCSDK
  • TinyYoloV2 for NCSDK
  • Inception for TensorFlow
• TensorFlow test for batch size fixed.
• Copyright license LGPL added.
• Supported platforms:
  • Intel Movidius Neural Compute Stick (version 1)
  • NVIDIA Jetson AGX Xavier
  • x86 systems
  • NVIDIA TX2
  • i.MX8

Known issues:

• NCSDK does not support multiple calls to the inference engine from the same thread. This causes the NCSDK backend to fail after the second start.

v0.2.0

Introduced features:

• Support for the following platforms:
  • Intel Movidius Neural Compute Stick (version 1)
  • NVIDIA Jetson AGX Xavier
  • x86 systems
  • NVIDIA TX2
  • i.MX8
• Compatibility support to handle models with generic or defined batch size

Known issues:

• NCSDK does not support multiple calls to the inference engine from the same thread. This causes the NCSDK backend to fail after the second start.

v0.1.0

Introduced features:

• Support for the following architectures:
  • GoogLeNet (Inception v4)
  • Tiny Yolo v2
• Support for the following backends:
  • NCSDK
  • TensorFlow
• Support for the following platforms:
  • Intel Movidius Neural Compute Stick (version 1)
  • NVIDIA Jetson AGX Xavier
  • x86 systems

Known issues:

• NCSDK does not support multiple calls to the inference engine from the same thread. This causes the NCSDK backend to fail after the second start.