Deep learning annotation training and prediction workflow for microscopy video data
Project description
VidTrain
Train deep neural networks to analyze video data.
Installation
- Install anaconda
- (Optional) Install a python-capable IDE like Visual Studio Code
- Open a command line terminal to install
vidtrain
:- Create a python 3.7 environment:
conda create --name vidtrain python=3.7
. Note, the python version must be exactly 3.7 [1] - Activate the environment
conda activate vidtrain
- Install vidtrain
pip install vidtrain
- Create a python 3.7 environment:
Run
Execute the following code in python:
import vidtrain
if __name__ == '__main__':
vidtrain.workflow.JunctionAnalysis().run()
Notes
[1] The code uses some features that were introduced in 3.7 (dictionaries that are ordered by default), meaning it will not work properly with python <3.7. Furthermore, it uses Tensorflow 1 , which dpes not support 3.8. Thus, the python version must be exactly 3.7. In the future, it is planned to migrate to Tensorflow 2 and at that point, vidtrain will also work on python 3.8. However, this will require rewriting at least the data generator for the validation data.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.