Biomedically motivated community-detection-based image segmentation
This small project utilizes the label propagation community detection algorithm in order to segment images. The initial abstraction process involves abstracting pixels to nodes and creating edges between nearby nodes that have similar luminosities. In order to combat over-segmentation, the resulting communities from the initial round of label propagation are then abstracted into the nodes for further rounds. The current results are not super great.
Requires Rust.
Installation is simple:
git clone https://github.com/wigasper/rusty-lung
cd rusty-lung
cargo build --release