Prototype tool for annotating images.
This package includes one CLI tool:
tator:
Usage: tator [OPTIONS] SRC DEST
Options:
-C, --overlay_color TEXT Color to use for overlay. Valid options are: 'b
lue','orange','green','red','purple','brown','p
ink','gray','olive','cyan'.Defualt: 'cyan'.
-A, --overlay_alpha FLOAT Transparency to use for overlay provided as
alpha value (0-1).Default: 0.3.
-w, --img_width INTEGER Width of src images in pixels. Default is 256
-h, --img_height INTEGER Height of src images in pixels. Default is 256
-W, --viewer_width INTEGER Width of viewer in pixels. Default is 325
-H, --viewer_height INTEGER Height of viewer in pixels. Default is 800
-f, --filetype TEXT File format for src images (as file extension).
Default is '.jpeg'
--help Show this message and exit.
Note: SRC and DEST should both be local directories. SRC should contain images to annotate, DEST will store results.
This utility provides a simple interface for performing image annotation, and specifically defining binary semantic segmentation.
- r/w from rasterio instead of skimage
- refactor to make code simpler, add docs
- add box selector feature?
- r/w from S3 directories as well as local dirs
- General requirements listed above
- Anaconda or Miniconda
This will install the s1_preprocessor package from the local repo in editable mode. Any changes to Python files within the local repo should immediately take effect in this environment.
-
Clone the repo
git clone https://github.com/GeoBigData/tatortot.git
-
Move into the local repo
cd tatortot
-
Create conda virtual environment
conda env create -f environment.yml
-
Activate the environment
conda activate tatortot
-
Install Python package
pip install -r requirements_dev.txt
- TBD