AquaPose
Measuring 3D fish posture and kinematics in a water tank is hard because the air-water interface refracts every camera ray, so ordinary multi-view triangulation is systematically wrong. AquaPose corrects for that: it takes synchronized multi-view video from a 12-camera aquarium rig and a refractive calibration and reconstructs per-fish 3D midlines across frames — arc-length-sampled anatomical keypoints triangulated and optionally spline-fitted — producing dense 3D trajectories and midline kinematics. It is built for behavioral and neuroscience researchers running multi-camera aquarium rigs who have video plus a calibration and want quantitative 3D posture out, as used for cichlid behavioral research.
Pipeline
AquaPose processes multi-view video through a 5-stage pipeline:
- Detection — YOLO-based fish detection (standard or oriented bounding boxes)
- Tracking — Per-camera 2D temporal tracking via OC-SORT
- Association — Cross-camera tracklet association using ray-ray geometry and Leiden clustering
- Midline — 2D midline extraction via YOLO-seg or YOLO-pose backends
- Reconstruction — DLT triangulation of 2D midlines into 3D B-spline midlines
Long videos are processed in fixed-size temporal chunks with identity continuity across chunk boundaries.
Quick Start
pip install aquapose
PyTorch is not pinned by this install; see the Installation page for the platform-specific PyTorch step before running the pipeline.
# Scaffold a project under ~/aquapose/projects/my_project
aquapose init my_project
# Run the pipeline against it
aquapose -p my_project run
To see it run end to end against real data, follow the
Tutorial,
which walks a full pipeline run against the published tutorial dataset
(10.5281/zenodo.22264079).
Documentation
Full documentation is at aquapose.readthedocs.io:
Citation
If you use AquaPose in your research, please cite the software via
CITATION.cff
or GitHub's "Cite this repository" button on the repository page. This is
distinct from citing the tutorial dataset
(10.5281/zenodo.22264079): cite the
software when you mean the AquaPose codebase, and cite the dataset when you
mean the sample data it ships with.
License
AquaPose is licensed under AGPL-3.0-or-later. See LICENSING.md for why.
2 maintainers
Metadata
2 maintainers
Release files for aquapose 4.0.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| aquapose-4.0.0.tar.gz | 813.8 kB | With provenance | Details for aquapose-4.0.0.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| aquapose-4.0.0-py3-none-any.whl | Python 3 | none | any | Details for aquapose-4.0.0-py3-none-any.whl |
Total release size: 1.2 MB
Release files / aquapose-4.0.0.tar.gz
| Download URL | aquapose-4.0.0.tar.gz |
|---|---|
| Size | 813.8 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
373ed87fa14554b6b23beb148a671e4308ca4a7baa7510459969b7788bb88960
|
|
BLAKE2b-256 checksum How to use checksums |
dea94a45e5b573c6cb6242ceb9a1415af744c226fea8883ef3c797607651f2e0
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 10, 2026.
Transparency logRelease files / aquapose-4.0.0-py3-none-any.whl
| Download URL | aquapose-4.0.0-py3-none-any.whl |
|---|---|
| Size | 421.2 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
9fed57710068645987cde9ad35a0a319324d8ed065afe81e498c5e5d21cda05c
|
|
BLAKE2b-256 checksum How to use checksums |
f20004ea5b1a89f957a26400b259d2d5dd17e5c969106815c12a26ad75e5a3e4
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 10, 2026.
Transparency logSecurity for aquapose 4.0.0
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
Full provenance
All 2 files in this release have been uploaded with provenance metadata.
Attestation for aquapose-4.0.0.tar.gz
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 10, 2026.
Transparency logA note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.
