Go language bindings for ONNX runtime
I'm a fan of Go and have just started digging a bit deeper in to machine learning. I heard about ONNX runtime and I'm a fan of standardization, so it seemed like a good place to start. I realized ONNX runtime didn't have Go language bindings, and I figured, if I can get that going, it'd probably be a great way to get started on my AI/ML journey.
The initial goal was to replicate the functionality of the C example from the ONNX repository, here.
At this point, the implemented functionality achieves the same result as the example noted above and I've, additionally, tested it with ResNet on image classification (example below and in main.go).
The API is incomplete (compared to the functionality available in the C library), at this time. I may try continue to build it out, as time permits, but would gladly accept help if anybody else is interested in this sort of thing.
Go-onnx uses cgo and leverages the onnxruntime shared library, so to run your program which leverages
go-onnx, you'll need to let cgo know where that library resides on your local system. To do so, in your main.go
(or wherever), include something like the following snippet:
/*
#cgo LDFLAGS: -L/path/to/onnx/runtime/lib
*/
import "C"The directory specified should contain the libonnxruntime.so (named the same). If your ONNX runtime file is named
something different, you may need to include the additional flag -l<libname>.
For a new application, first get go-onnx:
go get github.com/dhdanie/goonnx
You'll also need to download the example ResNet model from here.
Then, you should be able to run a basic demo application like the following (see main.go for working demo):
package main
/*
#cgo LDFLAGS: -L/usr/local/lib/onnx -lonnxruntime
*/
import "C"...
func classifyResNet(rgbVals []float32) [][]float32 {
defer timeTrack(time.Now(), "classifyResnet")
logId := "log0001"
var myCustomLogger ort.CustomLogger = func(severity ort.LoggingLevel, category string, codeLocation string, message string) {
fmt.Printf("Custom Logger %d/%s/%s - %s\n", severity, category, codeLocation, message)
}
env, err := ort.NewEnvironmentWithCustomLogger(ort.LoggingLevelVerbose, logId, myCustomLogger)
if err != nil {
errorAndExit(err)
}
defer env.ReleaseEnvironment()
opts := &ort.SessionOptions{
IntraOpNumThreads: 1,
GraphOptimizationLevel: ort.GraphOptLevelEnableBasic,
SessionLogID: logId,
LogVerbosityLevel: 0,
}
session, err := ort.NewSession(env, "models/resnet152v2.onnx", opts)
if err != nil {
errorAndExit(err)
}
defer session.ReleaseSession()
typeInfo, err := session.GetInputTypeInfo(0)
if err != nil {
errorAndExit(err)
}
tensorInfo, err := typeInfo.ToTensorInfo()
if err != nil {
errorAndExit(err)
}
memoryInfo, err := ort.NewCPUMemoryInfo(ort.AllocatorTypeArena, ort.MemTypeDefault)
if err != nil {
errorAndExit(err)
}
defer memoryInfo.ReleaseMemoryInfo()
value, err := ort.NewTensorWithFloatDataAsValue(memoryInfo, "data", rgbVals, tensorInfo)
if err != nil {
errorAndExit(err)
}
inputValues := []ort.Value{
value,
}
outs, err := session.Run(&ort.RunOptions{}, inputValues)
if err != nil {
errorAndExit(err)
}
outputs := make([][]float32, len(outs))
for i, out := range outs {
if out.GetName() != "resnetv27_dense0_fwd" {
continue
}
outFloats, err := out.GetTensorMutableFloatData()
if err != nil {
errorAndExit(err)
}
outputs[i] = make([]float32, len(outFloats))
for j := range outFloats {
outputs[i][j] = outFloats[j]
}
}
return outputs
}