R2Inference - EdgeTPU Examples

R2Inference
Introduction
Getting started
Getting the code Building the library
Supported backends
TensorFlow TensorFlow-Lite Caffe TensorRT Coral from Google ONNXRT ONNXRT ACL ONNXRT OpenVINO
Examples
TensorFlow TensorFlow-Lite TensorRT Coral from Google ONNXRT ONNXRT ACL ONNXRT OpenVINO
Model Zoo
Contact us

R2Inference provides several examples to highlight its usage with different architectures and frameworks. The models used for these examples can be downloaded from our Model Zoo.

Preparation

To test the EdgeTPU example you will need a Tensorflow-Lite C++ API and EdgeTPU API compatible graph file. You can obtain the model from the Model Zoo.

Inception v2

This example is located in r2inference/examples/r2i/edgetpu . To use this example run:

./inception -i [JPG input Image] -m [Tflite Model] -s [Model Input Size]

For example, evaluating this image:

Error creating thumbnail: Unable to save thumbnail to destination

Inception example input

Should produce the following output:

./inception -i Plane_example_input.jpg -m inception_v2_224_quant_edgetpu.tflite -s 224
Loading Model: inception_v2_224_quant_edgetpu.tflite
Setting model to engine
Loading image: Plane_example_input.jpg
Configuring frame
Starting engine
Predicting...
Highest probability is label 405 (0.730469)
Stopping engine

According to the ImageNet labels 405 corresponds to an 'airship'.