Code for Node Detection and Tracking in microscopic Phase Images for Cell Force Microscopy

This codebase expects input in either Phase image format or Fluroscent image format. If Phase images are used as input then a Pix2Pix model is used to generate fake fluroscent images and then the node detection and tracking network is used for tracking nodes to calucalute displacement and cell forces.

For training:

1. Train Pix2Pix using code from Pix2Pix repository, also placed in Pix2Pix/ Refer to tips in Pix2Pix/pytorch-CycleGAN-and-pix2pix/docs/tips.md
2. Node detection network with the train script in src/

For Inference:*

1. Run DataPrepforPix2Pix to prepare data in the format Pix2Pix expects for conversion from Phase data to Flurocent data. if phase data as input else start from step 3
2. Run generate_fluro.py to generate fluroscnet images from phase images using Pix2Pix trained model. update pix2pix_path, fake_fluro_path, phase_path and num_images (counted from 0 so 1+the actual last id you have)
3. Run DataPrep.ipynb for preparing fluroscent images for node tracking. Run DataPrep cell 5 if starting from step 1 or Run DataPrep cell 4 if starting from step 3 update video_name_list and master_Folder_name
4. Run detection_tracking.sh for node detection and tracking update video path in the loop definition