cellseg: Multiclass Cell SegmentationΒΆ
Version 0.1.0
Initialized script mode.
Optimized imports
Updated docs to reflect proper DataProcessor usage.
Initial tests.
DataProcessor
now errors if image and mask/label lengths differ.Dropped thresholding methods. Please use pyautocv or any other image processing package of your choice.
Refactored
show_images
to handle follow currentDataProcessor
logic. Fixed a bug that caused rows and column switch.DataProcessor
now returns a dictionary containing an image, its label, and index.Reading stacked tiff images is no longer supported for now.
dir_type
was dropped inDataProcessor
. Only provide a directory toimage_dir
containing images and masks.Added data from cytounet
Versioning is now automated.
Updated docs to show that this is a work in progress.
Updated docs to use explicit imports.
version 0.0.0
Preserve name on PyPI
Fixed issues with
show_images
showing blank images for masks (labels).Fixed issues with
uint16
not working withPIL
.DataProcessor
can now transform images to a given target size.Renamed
DataLoader
class toDataProcessor
to avoid conflicts withtorch.utils.data.DataLoader
Added initial simple CNN model with a single layer
Added
show_images
inutils.py
to allow quick visualization of a given number of images from a given stack of images.Implemented data loaders.
Conceptualized project