Class Net

Constructors

constructor

Methods

addLayer addLayerToPrev connect dump dumpToFile empty enableFusion forward forwardAsync getLayerNames getLayerNamesAsync getPerfProfile getUnconnectedOutLayers getUnconnectedOutLayersAsync setInput setInputAsync setPreferableBackend setPreferableTarget

Constructors

constructor

Methods

addLayer

  • addLayer(name, type, params): number

  • Parameters

    Returns number

addLayerToPrev

  • addLayerToPrev(name, type, params): number

  • Parameters

    Returns number

connect

  • connect(outPin, inpPin): void

  • Parameters

    • outPin: string
    • inpPin: string

    Returns void

  • connect(outLayerId, outNum, inpLayerId, inpNum): void

  • Parameters

    • outLayerId: number
    • outNum: number
    • inpLayerId: number
    • inpNum: number

    Returns void

dump

  • dump(): string

  • Returns string

dumpToFile

  • dumpToFile(path): void

  • Parameters

    • path: string

    Returns void

empty

  • empty(): void

  • Returns void

enableFusion

  • enableFusion(fusion): void

  • Parameters

    • fusion: boolean

    Returns void

forward

forwardAsync

  • forwardAsync(inputName?): Promise<Mat>

  • Parameters

    • Optional inputName: string

    Returns Promise<Mat>

  • forwardAsync(outBlobNames?): Promise<Mat[]>

  • Parameters

    • Optional outBlobNames: string[]

    Returns Promise<Mat[]>

getLayerNames

  • getLayerNames(): string[]

  • Returns string[]

getLayerNamesAsync

  • getLayerNamesAsync(): Promise<string[]>

  • Returns Promise<string[]>

getPerfProfile

  • getPerfProfile(): {
        retval: number;
        timings: number[];
    }

  • Returns overall time for inference and timings (in ticks) for layers.

    Indexes in returned vector correspond to layers ids. Some layers can be fused with others, in this case zero ticks count will be return for that skipped layers. Supported by DNN_BACKEND_OPENCV on DNN_TARGET_CPU only.

    https://docs.opencv.org/4.x/db/d30/classcv_1_1dnn_1_1Net.html#a06ce946f675f75d1c020c5ddbc78aedc

    [out] timings vector for tick timings for all layers. Returns overall ticks for model inference. WARN retval is a int64, which can overflow nodejs number

    Returns {
        retval: number;
        timings: number[];
    }

    • retval: number
    • timings: number[]

getUnconnectedOutLayers

  • getUnconnectedOutLayers(): number[]

  • Returns indexes of layers with unconnected outputs.

    FIXIT: Rework API to registerOutput() approach, deprecate this call

    Returns number[]

getUnconnectedOutLayersAsync

  • getUnconnectedOutLayersAsync(): Promise<number[]>

  • Returns Promise<number[]>

setInput

  • setInput(blob, name?, scalefactor?, mean?): void

  • Sets the new input value for the network.

    https://docs.opencv.org/4.x/db/d30/classcv_1_1dnn_1_1Net.html#a5e74adacffd6aa53d56046581de7fcbd

    Parameters

    • blob: Mat

      A new blob. Should have CV_32F or CV_8U depth.

    • Optional name: string

      A name of input layer.

    • Optional scalefactor: number

      An optional normalization scale.

    • Optional mean: number

      An optional mean subtraction values.

    Returns void

  • setInput(blob, inputName?): void

  • Parameters

    • blob: Mat
    • Optional inputName: string

    Returns void

setInputAsync

  • setInputAsync(blob, inputName?): Promise<void>

  • Parameters

    • blob: Mat
    • Optional inputName: string

    Returns Promise<void>

setPreferableBackend

  • setPreferableBackend(backendId): void

  • Ask network to use specific computation backend where it supported.

    Parameters

    • backendId: number

      backend identifier.

    Returns void

setPreferableTarget

  • setPreferableTarget(targetId): void

  • Ask network to make computations on specific target device.

    Parameters

    • targetId: number

      target identifier.

    Returns void

Settings

Member Visibility

Theme

On This Page