/EddyData

Keras implementation based on Tensorflow and dataset SCSE-Eddy for article "A Deep Framework for Eddy Detection and Tracking from Satellite Sea Surface Height Data"

Primary LanguagePython

EddyData

Keras implementation based on Tensorflow and dataset SCSE-Eddy for article "A Deep Framework for Eddy Detection and Tracking from Satellite Sea Surface Height Data"

The Structure of Our FrameWork

structure.jpg

Example for Detection Result

result.jpg

Download the Dataset SCSE-Eddy

Unzip and locate it

Train and Validate

python3 FrameWork.py

store the detection results on your own path

Example for Tracking Result

track.jpg

Tracking

load the detection results and then

python3 trace.py

Comparison of Number of Eddies Detected by Different Methods

Methods N-Cyclone N-Anticyclone N-All
PET 4437 3737 8210
EddyNet 4502 3794 8296
PSPNet 4614 3906 8520
Ours 4662 3952 8614

Others

Please refer to the citations in this paper(access address: https://doi.org/10.1109/TGRS.2020.3032523 or https://ieeexplore.ieee.org/document/9247537) for SCSE-Eddy dataset. We will continuously update for researchers.

We look forward to new breakthroughs for scholars in the future !