Crate deoldify 

Structs

Error

Constants

MODEL_SIZE

Functions

num_cpus

postprocess

Convert model output back to an image, merging original luminance with predicted chrominance.

prepare_tensor

Resize grayscale image to MODEL_SIZE square and convert to NCHW f32 tensor (values 0-255).

preprocess_image

Load the input image, extract original L (luminance) channel, and create 3-channel grayscale.

run_inference

Run ONNX inference on the input tensor and return the output tensor.

Type Aliases

Result