Fast labeling with MIA utilizing segment anything
MIA is a software for deep learning based image analysis. It covers image labeling, neural network training and inference. It can be used for image classification, object detection, semantic segmentation and tracking.
The easiest way to install MIA is via conda (see https://docs.conda.io/en/latest/miniconda.html for installation options).
After installation of conda, download the environment file.
Then, open an anaconda prompt and type:
cd /path/to/mia_environment.yaml(change/path/to/to the path of the folder with the environment file)conda env create -f mia_environment.yaml- wait and follow instructions
type in an anaconda prompt:
conda activate mia_environmentmianalyzer
A quickstart guide can be found here and the complete user manual here.
Please use image.sc with the mia-tag for general discussion, questions about how to use the software or feature requests. Bugs can be reported directly in the issues panel on github.
If you use this code for your research, please cite:
https://www.cell.com/cell-reports-methods/pdf/S2667-2375(23)00146-7.pdf
Körber, MIA is an open-source standalone deep learning application for microscopic image analysis, Cell Reports Methods (2023)
In general, MIA should run on any system with Linux or windows. You can use the cpu only, but it is highly recommended to have a system with a cuda-compatible gpu (from NVIDIA) to accelerate neural network training.