Class InferenceHandler

Note

Construct via the Class AniraWeb factory in most cases: aniraWeb.InferenceHandler(...) rather than new InferenceHandler(...). The factory threads the WASM instance through for you.

class InferenceHandler(wasmInstance, preprocessor, config, customProcessor)

TypeScript wrapper for anira::InferenceHandler Thread-safe C API wrapper

Extends:

  • BaseWrapper

Arguments:

  • wasmInstance (AniraWasmInstance)

  • preprocessor (PossiblePointer<PrePostProcessor>)

  • config (PossiblePointer<InferenceConfig>)

  • customProcessor (PossiblePointer)

destroy()

Free the underlying C++ object. See Lifecycle and Cleanup for when to call this.

getAvailableSamples(tensorIndex, channel=0)

Mirrors anira::InferenceHandler::get_available_samples().

Arguments:

  • tensorIndex (number)

  • channel (number)

Returns:

number

getInferenceBackend()

Mirrors anira::InferenceHandler::get_inference_backend().

Returns:

number

getLatency(tensorIndex=0)

Mirrors anira::InferenceHandler::get_latency().

Arguments:

  • tensorIndex (number)

Returns:

number

getLatencyVector()

Mirrors anira::InferenceHandler::get_latency_vector().

Returns:

number[]

popData(outputPtr, numSamples, tensorIndex=0)

Mirrors anira::InferenceHandler::pop_data() (non-blocking).

Arguments:

  • outputPtr (number)

  • numSamples (number)

  • tensorIndex (number)

Returns:

number

popDataBlocking(outputPtr, numSamples, waitMs, tensorIndex=0)

Mirrors anira::InferenceHandler::pop_data() (blocking with timeout).

Arguments:

  • outputPtr (number)

  • numSamples (number)

  • waitMs (number)

  • tensorIndex (number)

Returns:

number

popDataMulti(outputPtr, numSamplesPtr)

Mirrors anira::InferenceHandler::pop_data() (multi-tensor, non-blocking).

Arguments:

  • outputPtr (number)

  • numSamplesPtr (number)

Returns:

number

popDataMultiBlocking(outputPtr, numSamplesPtr, waitMs)

Mirrors anira::InferenceHandler::pop_data() (multi-tensor, blocking with timeout).

Arguments:

  • outputPtr (number)

  • numSamplesPtr (number)

  • waitMs (number)

Returns:

number

prepare(hostConfig)

Mirrors anira::InferenceHandler::prepare().

Arguments:

  • hostConfig (PossiblePointer<HostConfig>)

prepare(hostConfig, customLatency, tensorIndex)

Mirrors anira::InferenceHandler::prepare().

Arguments:

  • hostConfig (PossiblePointer<HostConfig>)

  • customLatency (number)

  • tensorIndex (number)

prepare(hostConfig, customLatency)

Mirrors anira::InferenceHandler::prepare().

Arguments:

  • hostConfig (PossiblePointer<HostConfig>)

  • customLatency (Uint32Array)

process(dataPtr, numSamples, tensorIndex)

Mirrors anira::InferenceHandler::process().

Arguments:

  • dataPtr (number)

  • numSamples (number)

  • tensorIndex (number)

Returns:

number

process(inputPtr, numInputSamples, outputPtr, numOutputSamples, tensorIndex)

Mirrors anira::InferenceHandler::process().

Arguments:

  • inputPtr (number)

  • numInputSamples (number)

  • outputPtr (number)

  • numOutputSamples (number)

  • tensorIndex (number)

Returns:

number

processMulti(inputPtr, numInputPtr, outputPtr, numOutputPtr)

Mirrors anira::InferenceHandler::process() (multi-tensor overload).

Arguments:

  • inputPtr (number)

  • numInputPtr (number)

  • outputPtr (number)

  • numOutputPtr (number)

Returns:

number

pushData(inputPtr, numSamples, tensorIndex=0)

Mirrors anira::InferenceHandler::push_data().

Arguments:

  • inputPtr (number)

  • numSamples (number)

  • tensorIndex (number)

pushDataMulti(inputPtr, numSamplesPtr)

Mirrors anira::InferenceHandler::push_data() (multi-tensor overload).

Arguments:

  • inputPtr (number)

  • numSamplesPtr (number)

reset()

Mirrors anira::InferenceHandler::reset().

setInferenceBackend(backend)

Mirrors anira::InferenceHandler::set_inference_backend().

Arguments:

  • backend (number)

setNonRealtime(nonRealtime)

Mirrors anira::InferenceHandler::set_non_realtime().

Arguments:

  • nonRealtime (boolean)