Read a NumPy .npy integer mask

Read a NumPy integer array (e.g. a ground-truth annotation mask) into a lossless annot_mask – values are preserved exactly, including composite bitfield codes. Use this (not at_read_mask(), which polygonises) when the raw integer matrix is what you need, for example to derive a training mask or compute agreement against another mask.

Usage

at_read_npy(
  path,
  level = 0L,
  transpose = FALSE,
  legend = NULL,
  call = rlang::caller_env()
)

Arguments

path

Path to a .npy file.

level

Integer pyramid level to record. Default 0.

transpose

Logical; transpose the array on read, for files stored in the native cube order (width, height) = [x, y] (e.g. TIVITA/HyperGui masks) so the result lands in annotatR's [y, x] convention. Default FALSE assumes the standard image convention (height, width).

legend

Optional legend tibble (value, label) overriding the sidecar / pixel-value labelling.

call

The calling environment, for error reporting.

Value

An annot_mask.

See also

at_write_npy(), at_read_mask()

Other masks: at_mask(), at_mask_agreement(), at_mask_boundary(), at_mask_derive(), at_mask_legend(), at_mask_preview(), at_mask_stack(), at_mask_stats(), at_read_mask(), at_write_mask(), at_write_masks(), at_write_npy()

Examples

m <- at_mask(at_example_project(), "labelled")
f <- tempfile(fileext = ".npy")
at_write_npy(m, f)
at_read_npy(f)
#> <annot_mask> labelled  |  512 x 512 px  |  level 0
#> values: 3
#> # A tibble: 3 × 6
#>   value label    layer   roi_id         n_px colour 
#>   <int> <chr>    <chr>   <chr>         <int> <chr>  
#> 1     1 tumour   regions roi_000000049 19600 #999999
#> 2     2 necrosis regions roi_000000050 72704 #E69F00
#> 3     3 stroma   regions roi_000000051 34200 #56B4E9