/Rheotaxis-zebrafish

Atomated zebrafish rheotactic behavior analysis platform.

Primary LanguagePythonMIT LicenseMIT

Rheotaxis-zebrafish

Automated zebrafish reheotaxis analysis platform.

Consists:

  1. library for zebrafish detection and tracking (zebrafish)
  2. interactive app (zebrafish_on_jupyter)

Interactive app runs on top of nteract

Installation

Full installation (requires pip):

  1. Install zebrafish library:
pip install -e .
  1. Install zebrafish_on_jupyter for interactive app:
cd jupyter-extension-zebrafish/
pip install -e .
jupyter serverextension enable --py zebrafish_on_jupyter --sys-prefix

Usage

  • Run(in folder containing app_notebook.ipynb):
jupyter zebrafish.

User interface will pop up using a browser. (If not, copy and paste the notebook URL from the terminal.)

Front

  • Before running Detection or Postprocess tab, Initialize by clicking Initialize button. INFO will be printed below if initialized without errors.

    Initialize

  • Put files to process at data/ and click Get Current Files button to get file names.

  • Detection file paths: gets all files with mp4 extension.

  • Post processing excel file paths: gets all excel files with names that ends with _detection.

  • Post processing video file paths: gets all excel files with names that ends with _detection_post and gets videos(mp4) with same names except _detection_post.

  • Excluded file paths: any other files.

  • Detection preview file: first mp4 file from Detection file paths. Used for Preview Detection.

GetFiles

1. Notebook

Notebook(app_notebook.ipynb) consists python code for running user's frontend.

2. Detection

  • Output Type:

    • excel: Outputs detection data in excel from videos in Detection file paths.
    • video: Outputs detection video from videos in Detection file paths.
    • excel+video: Both from above.

  • Detection Method:

    • ThresholdDetector: Detector using binary thresholding.
      • Image Crop
        • Height bound: Overall crop height bound.
        • Width bound: Overall crop width bound.
      • Flow
        • Check if water flows left to right: Used for calculating relative angle of zebrafish to water flow.
      • Thresholds
        • Median Blur Size: Median blur kernel size. Median blur is applied on l_channel image.
        • L Channel Threshold: Binary threshold.
        • Size bound: Size bound of zebrafish in pixels.
        • Closing iterations: Iterations for dilation and erosion (closing).

  • Detection preview: If current settings are saved, you can see its results by clicking Show Preview button.

    Preview

  • Run detection by clicking Run button or interrupt by Interrupt button.

    Run

3. Postprocess

  • Output Type:

    • excel: Outputs post processed data in excel from excel files in Post processing excel file paths. (csv files are also generated.)
    • video: Outputs post processed video from excel files and matching video files in Post processing video file paths.

  • Tracker:

    • KalmanTracker: Tracking by multi-object kalman filter.
      • Max age: maximum age for tracked object.
      • Min hits: minimum threshold of detections for a new object.
      • Distance Threshold: maximum distance threshold for updating existing object.
    • VicinityTracker: Tracking by searching vicinity.
      • Number of frames to find: max number of frames for searching.
      • Max distance to find: max distance for searching in pixels.
      • Max distance increment ratio per frame: max distance is incremented per frame.

  • Methods:

    • DataConverter:
      • Frames per second: fps for calculating time related values(velocity ...).
      • Pixel to cm ratio: Used for converting pixel measurements to cm.
      • Converting methods: select converting methods.
    • Position bounder: add position bounds for checking if position is in the added bounds.(Bool)
      • Current Bounds: shows current added bounds.