Developed as part of the Digital Typhoon project from Kitamoto-sensei. Provides a set of tools to enable easy and pythonic interaction the Digital Typhoon dataset.
Full documentation here.
| Section | Description |
|---|---|
| pyphoon | Library for Digital Typhoon project |
| docs | Library documentation files |
| notebooks | Basic code examples. (will be removed in near future) |
| scripts | Some example scripts using library tools |
| sampledata | Sample data from Digital Typhoon, used in |
| experiments | Data and files related to specific applications of pyphoon library (includes notebooks). |
Refer to the instructions here.
# Load a sequence
from pyphoon.io.h5 import read_source_images
from pyphoon.io.utils import get_image_ids
images = read_source_images('sampledata/datasets/image/200717')
images_ids = get_image_ids('sampledata/datasets/image/200717')
# Display sequence
from pyphoon.visualise import DisplaySequence
DisplaySequence(
images=images,
images_ids=images_ids,
name='200717',
interval=100
).run()
pyphoon was mainly conceived to assist researchers in Machine Learning/Deep Learning experiments. To this end, this repository provides examples of experiments carried by Kitamoto-lab interns:
| Section | Description |
|---|---|
| tcxtc | Tropical cyclone vs Extratropical cyclone binary classifier. |
| multiclass | Classification of Topical cyclone intensity in four categories. |
| pressure regression | Regression of the centre pressure in Tropical cyclones |
Note: All models have been implemented using keras.