Crate tfdeploy

Tensorflow Deploy

Tiny, no-nonsense, self contained, portable Tensorflow inference.

Example

use tfdeploy::*;

// load a simple model that just add 3 to each input component
let model = tfdeploy::tf::for_path("tests/models/plus3.pb").unwrap();

// "input" and "output" are tensorflow graph node names.
// we build an execution plan for computing output from input
let plan = SimplePlan::new(&model, &["input"], &["output"]).unwrap();

// run the computation.
let input = ndarray::arr1(&[1.0f32, 2.5, 5.0]);
let mut outputs = plan.run(tvec![input.into()]).unwrap();

// take the tensors coming out of the only output node
let mut tensors = outputs.pop().unwrap();

// grab the first (and only) tensor of this output, and unwrap it as array of f32
let tensor = tensors.pop().unwrap().take_f32s().unwrap();
assert_eq!(tensor, ndarray::arr1(&[4.0, 5.5, 8.0]).into_dyn());

For a more serious example, see inception v3 example.

Re-exports

pub use errors::*;

pub use analyser::TensorFact;

pub use dim::TDim;

pub use model::Model;

pub use model::Node;

pub use model::TVec;

pub use plan::SimplePlan;

pub use tensor::DatumType;

pub use tensor::Tensor;

pub use model::tf;

Modules

analyser
dim
errors	error_chain generated types
macros
model
ops	TensorFlow Ops
plan
streaming
tensor	Tensor is the equivalent of Tensorflow Tensor.
tfpb	Generated protobuf codec for Tensorflow models, plus a handful of helper for writting tests.

Macros

dimfact	Constructs a dimension fact.
map_tensor
shapefact	Constructs a shape fact.
tvec
typefact	Constructs a type fact.
unwrap_or_none	Tries to unwrap an option, or returns Ok(None) otherwise.
valuefact	Constructs an value fact.
wrap