Automl VR - watch on youtube
An end-to-end process to build and deploy an object detection model that runs within VR using Google Cloud AutoML model exported to the edge.
https://www.github.com/tensorflow/tensorflow/tree/master/tensorflow/lite/experimental/examples/unity/TensorFlowLitePlugin (don't bother compiling your own TFLite library, use the pre-built ones in this repo)
https://gist.github.com/ZackAkil/ce6604fd5ac008756f938047ce73d9d5
https://gist.github.com/ZackAkil/b22471aefa05fac6ad46f7589081dffb
writer = new StreamWriter(csv_path, true);
// create label row for Google AutoML Object Detection
string row = string.Format("UNASSIGNED,{0}{1},{2}, {3}, {4},,,{5},{6},,", "BUCKET_LOCATION",
file_name,
label_name,
x_relative_min,
y_relative_min,
x_relative_max,
y_relative_max);
writer.WriteLine(row);