Examples#

Worked walkthroughs on real data. The Guide explains how each part of sidescanpy works; these pages show it doing a job, and Theory explains the physics behind the steps.

Each example below runs sidescanpy on a real, openly licensed recording that anyone can download, and shows the code, what it prints and the figure it makes. The pages are built by running the scripts in examples/gallery/. Another section sits alongside them:

  • Target recognition: two-stage target recognition step by step, from raw data or a target catalog to a trained detector and classifier. Those scripts need a GPU and extra packages, so they are not run when this site is built; each page shows the results of a run.

Run one yourself with an output directory (the current one by default):

python examples/gallery/plot_correct_a_line.py out/

Plotting needs the viz extra and the GeoTIFF in Grid to a map the geo extra. The recordings, their sizes and licences are listed in Example data.

Read a file

Read a file

What’s in a file: metadata and aux streams

What's in a file: metadata and aux streams

Navigation: held fixes, interpolation and the track

Navigation: held fixes, interpolation and the track

Investigating an unknown survey

Investigating an unknown survey

Correct a line

Correct a line

Comparing results across sonars and file formats

Comparing results across sonars and file formats

Grid to a map

Grid to a map

Eight wrecks at one scale

Eight wrecks at one scale

Four passes over one wreck: native and north-up

Four passes over one wreck: native and north-up

Does an object look the same from every pass?

Does an object look the same from every pass?

Swath tiles for annotation, and boxes back to targets

Swath tiles for annotation, and boxes back to targets

Target chips at several corrections and resolutions

Target chips at several corrections and resolutions

A training dataset with background images

A training dataset with background images

Same pings, different chains: recreating a vendor product

Same pings, different chains: recreating a vendor product

Four published chains, four looks; one recorded chain, one scale

Four published chains, four looks; one recorded chain, one scale