Description:
reviewer2 0.9.31
reviewer #2
This tool starts a simple image server that lets you quickly flip through image files from a local directory using your web browser.
Also, it optionally shows a customizable form where you can take notes or answer questions about each image or set of images.
Example uses:
manual curation / review of sequencing data visualization images such as those generated by REViewer for short tandem repeat loci
machine learning training set creation
reviewing a pile of photos
Features:
simple way to flip through many local image files using your web browser
crawls a top-level directory to find .png, .jpeg, or .svg image files
web interface: home page lists all images
web interface: image pages show the image, an optional customizable form where you can take notes or answer questions about the image, next/previous page links, and optional other customizable info for context
use subdirectories to group images. Any images found in the same subdirectory will be shown on the same page
Install:
python3 -m pip install reviewer2
Run:
python3 -m reviewer2 # start server for all images in the current directory and subdirectories
Below are more examples. Run with --help to see all available options and their descriptions.
python3 -m reviewer2 -x temp -x keyword2 /path/dir-with-images # -x are keyword(s) of paths to skip and /path/dir-with-images is the top level dir to search instead of the current dir
python3 -m reviewer2 -t /path/user_responses.xls # change where user responses get saved (default: reviewer2_form_responses.tsv)
python3 -m reviewer2 -m /path/metadata.tsv # provide a metadata table
After the server is running, open your web browser to http://localhost:8080 to start reviewing images.
Options:
metadata table (-m)
It's often useful to add extra info to the image pages to help with review - such as image descriptions, quality scores, etc.
To enable this, there are several ways to specify arbitrary key-value pairs to add to specific image pages.
The 1st way is to put a file called reviewer2_metadata.json next to the image(s). All keys and values from this file
will appear on that image page. The 2nd way is to use -m to pass in a metadata table
(.tsv or .xls) with a Path column + arbitrary other columns. If the Path value matches the relative directory containing
the image(s), entries from that row will be added to this image page.
Since the keys and values are treated as html, they can be used to add more complex info - such as
colors, text formatting, <img ..> tags with images from other web pages, iframes containing entire sections of external pages, etc.
responses table (-t)
As you fill in the forms at the top of the image pages, the responses are written to this table. If you later restart reviewer2 with the same -t, it will reload previous responses. You can also optionally use this table to provide additional columns to display - sometimes this can be more convenient than using -m.
Default: reviewer2_form_responses.tsv
custom form schema (--form-schema-json)
If you'd like to use non-default questions in the image page forms, you can specify the path or url of a .json file containing a custom form schema. For examples of the expected format see main/form_schema_examples
config file (~/.reviewer2_config)
Most settings that can be provided on the command line can also be set via this YAML config file instead. For example:
~/.reviewer2_config
form-schema-json: /path/to/my-schema.json
hide-metadata-on-home-page: true
host: 127.0.0.1
port: 8080
For more details, run:
python3 -m reviewer2 --help
Comparing reviews:
If 2 or more people review the same images, the reviews can be compared using the compare_form_response_tables script.
For example:
python3 -m compare_form_response_tables egor_reviewer2_form_responses.tsv ben_reviewer2_form_responses.tsv -o combined_responses.tsv -s1 egor -s2 ben
This script will print concordance stats, and output a combined_responses.tsv table that contains one image per row as well as each person's review of the image.
To see the full list of args and descriptions, run python3 -m compare_form_response_tables --help
Development:
To create a local dev instance, run
git clone git@github.com:broadinstitute/reviewer2.git
cd reviewer2
# start server in dev mode so it reloads code on change
python3 -m reviewer2 /path/dir-with-images --dev-mode
License
For personal and professional use. You cannot resell or redistribute these repositories in their original state.
Customer Reviews
There are no reviews.
