Semantic labeling made simple
Project description
Semi-Supervised Semantic Annotator (S3A)
A highly adaptable tool for both visualizing and generating semantic annotations for generic images.
Most software solutions for semantic (pixel-level) labeling are designed for low-resolution (<10MB) images with fewer than 10 components of interest. Violating either constraint (e.g. using a high-res image or annotating ~1000 components) incur detrimental performance impacts. S3A is designed to combat both these deficiencies. With images up to 150 MB and 2000 components, the tool remains interactive.
A more detailed overview can be found in the project wiki here.
Installation
The easiest method for installing s3a
is via pip
after cloning the repository, or directly from pypi:
git clone https://gitlab.com/ficsresearch/s3a
pip install -e ./s3a
# Or from pypi using "pip install s3a"
Note that a version of OpenCV and Qt binding are required for S3A to work. These can be installed for you with the "turnkey" option:
pip install -e ./s3a[full]
# Or "pip install s3a[full]"
Running the App
Running the app is as easy as calling s3a
as a module:
python -m s3a
From here, projects can be created to host groups of related images, or images can be annotated in the default project. Both options are available through the File
menu.
Detailed Feature List
More information about the capabilities of this tool are outlined in the project wiki.
License
This tool is free for personal and commercial use (except the limits imposed by the PyQt5 library). If you publish something based on results obtained through this app, please cite the following papers:
Jessurun, N., Paradis, O., Roberts, A., & Asadizanjani, N. (2020). Component Detection and Evaluation Framework (CDEF): A Semantic Annotation Tool. Microscopy and Microanalysis, 1-5. doi:10.1017/S1431927620018243
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.