FED-PV: A large scale synthetic frame/event dataset for particle-based velocimetry.

Description

This is code accompanying the dataset and paper by Xiang Feng, Fan Wu and Aoyu Zhang.

Dataset Presentation

The following are the six flow fields in our dataset.

Each data item contains four grayscale particle images, 15 ms of event data, and the corresponding ground-truth motion field. For example:

\cylinder\cylinder_Re150_00001_flow-PIV-0.png \cylinder\cylinder_Re150_00001_flow-PIV-1.png \cylinder\cylinder_Re150_00001_flow-PIV-2.png \cylinder\cylinder_Re150_00001_flow-PIV3.png \cylinder\cylinder_Re150_00001_flow.h5 \cylinder\cylinder_Re150_00001_flow.flo

Install

The dataset is publicly available on the OneDrive platform:

URL: https://1drv.ms/f/c/b5123d043eaaea38/Esyx_kKJJaVNv3AgLJzcISEBGLji58gVUJqrTs4O9OAhog?e=161kuT Password: FEDPV666

Directory structure

main   
├─ flow_sets
│  └─ [flow field name]   
│     └─ *.flo 
├─ dataset   
│  └─ [flow field name]   
│     └─ *.png, *.h5   
├─ evt   
│  └─ [flow field name] 
│     └─ *.pnd, *.evt   
├─ utils   
│  ├─ class_dict.py  
│  ├─ plot.py  
│  ├─ progress_bar.py  
│  └─ smoothn.py  
├─ gen_event.py  
├─ gen_particle.py   
├─ main_dataset.py
├─ read_flow.py 
└─ save_file.py