VesicleYOLO-pentadec 0.0.2

A simple package created to easily apply YOLO-based vesicle detection to fluorescence microscopy images.

VesicleYOLO

This is a simple package meant to make the application of YOLO to vesicle quantification as easy as possible.

Installation

This is really easy. An example of the code is shown below.

pip install VesicleYOLO-pentadec

Example

from VesicleYOLO_pentadec.analyze import predict_folder

# A list of vesicles from the folder 'example_folder', with annotations passed into 'package_out'. 
vesicle_list = predict_folder("weights.pt",
                              r"example_folder",
                              "package_out")

Keep in mind predict_folder is for batch image processing and will return a list of image names alongside vesicle counts. predict_image is for singular images and will return a count only.

Notes on Parameters - what you can do with VesicleYOLO

predict_folder takes the following inputs:

weights_path: Currently required, the path to the model weights. In the future (tomorrow), I'm going to implement default weights.

images_directory: Required, the path to the images to be analyzed.

out_directory: Not required. Where annotations will be written; if annotations are chosen but no out_directory is set, a new directory is written. This is pretty easy but I haven't checked if it works yet haha.

num_slices: Not required, default is 256. Number of slices to cut the images into. Probably don't mess with this.

annotate_rect: Whether to draw rectangular 'bounding box' annotations in the out directory. Default is True.

annotate_ellipse: Whether to draw elliptical annotations in the out directory. Default is False.

predict_image takes the same inputs except for these things which you should leave blank:

length_divider: Amount to divide each length (height and width) by. I can't think of a conceivable reason why you would change this.

model_weights: Also not really something you should change takes the model weights from predict_folder.