classifier

Documentation

Index

• func Train(weights string, dataset dataset.Dataset) error
• type Inference
  • func NewInference(weights string) (*Inference, error)
  • func (in *Inference) Predict(image []byte) (int, float64, error)

Functions

func Train

func Train(weights string, dataset dataset.Dataset) error

Train will train a mnist network for given dataset object.

Types

type Inference

type Inference struct {
	// contains filtered or unexported fields
}

Inference is the internal representation of the inference object.

func NewInference

func NewInference(weights string) (*Inference, error)

NewInference will return a new inference object with the given trained weights file to handle predictions.

func (*Inference) Predict

func (in *Inference) Predict(image []byte) (int, float64, error)

Predict will take a 28x28 image and return the integer it thinks best matches the image. If no match is found, it will return -1.