Point Proofread

Proofreading point annotations on images with napari

  • Seongbin Lim

https://github.com/LaboratoryOpticsBiosciences/napari-point-proofread

License MIT PyPI Python Version tests codecov napari hub npe2 Copier

Proofreading point annotations on images with napari


[!IMPORTANT] Since v0.4.0 this plugin is published as napari-point-proofread. It was previously released as napari-proofread-brainbow (up to v0.3.3), which is no longer maintained. If you have the old package installed, run pip uninstall napari-proofread-brainbow before installing the new one.

This napari plugin was generated with copier using the napari-plugin-template.

Installation

You can install napari-point-proofread via pip:

pip install napari-point-proofread

If napari is not already installed, you can install napari-point-proofread with napari and Qt via:

pip install "napari-point-proofread[all]"

Beginner Quick Guide (Step-by-Step)

This plugin provides two panels in napari:

  • Plugins > Point Proofread > Manual Proofread
  • Plugins > Point Proofread > Proofread Predictions (from csv)

1. Open napari and load your data

  1. Start napari:

    napari
    
  2. Drag and drop your image into the viewer (or use File > Open File(s)...).

  3. If you have prediction points, also load the CSV file.

Screenshot placeholder: napari with raw image and optional CSV loaded

2. Open the plugin widgets

  1. Open Plugins > Point Proofread > Manual Proofread.
  2. (Optional) Open Plugins > Point Proofread > Proofread Predictions (from csv) if you want to review model predictions.

Screenshot placeholder: plugin menu showing both widgets

3. Prepare the image for proofreading

In Manual Proofread:

  1. Select your image in Image layer.
  2. Click Convert to RGB (for 4D channel-first/channel-middle data).
  3. Click Normalize (perc=99) to improve visibility.
  4. Use Contrast max and ZYX scale sliders as needed.
  5. (Optional) Click Make grid to create a labeled grid overlay for systematic review.

Tip: In napari layer list, right-click an RGB layer and use Split RGB when needed (often helpful for 3D inspection).

Screenshot placeholder: Manual Proofread panel and adjusted image view

4. Add or edit proofreading points

  1. Create or select a Points layer.
  2. In Point Layer Tools, set Point size for better visibility.
  3. (Optional) Enable Enable class annotation.
  4. Choose a class number and use Assign class to selected to label points.

Screenshot placeholder: points editing and class annotation controls

5. Review prediction CSV points by confidence (optional)

In Proofread Predictions (from csv):

  1. Select the loaded prediction points layer.
  2. Adjust Min probability.
  3. Click Threshold probability (csv) to create/update a filtered layer (threshold_prob).

Screenshot placeholder: thresholded prediction points layer

6. Save your corrected points

  1. Select your final points layer.
  2. Save as CSV from napari.
  3. In the save dialog, set Files of type to Point Proofread (*.csv) (as shown in the image). Otherwise, values that should remain integers (for example, class labels) may not be saved correctly.

Screenshot placeholder: saving points layer to CSV

Developer Setup

After cloning the repository, you can set up a development environment with conda:

pip

Step 1: Create and activate the corresponding conda environment

conda create -n napari-point-proofread -c conda-forge python=3.11 pip -y
conda activate napari-point-proofread

Step 2: Install requirements with pip

cd napari-point-proofread
pip install -e . --group dev

--group requires pip >= 25.1.

Step 3: Test the plugin by running napari

napari

The plugin should appear under Plugins > Point Proofread

With uv

uv sync
uv run napari

Run the tests with uv run pytest (current environment) or uv run tox (full matrix).

Contributing

Contributions are very welcome. Tests can be run with tox, please ensure the coverage at least stays the same before you submit a pull request.

License

Distributed under the terms of the MIT license, "napari-point-proofread" is free and open source software

Issues

If you encounter any problems, please [file an issue] along with a detailed description.

Version:

  • 0.4.0

Last updated:

  • 2026-11-04

First released:

  • 2026-11-04

License:

  • MIT

Supported data:

  • Information not submitted

Plugin type:

Open extension:

Save extension:

Python versions supported:

Operating system:

  • Information not submitted

Requirements:

  • napari
  • numpy
  • magicgui
  • qtpy
  • napari[all]; extra == "all"
Website by the napari team, original design by CZI.