metadata
license: mit
dataset_info:
features:
- name: image
dtype: image
- name: time
dtype: timestamp[us, tz=-05:00]
- name: ghi
dtype: float32
- name: dni
dtype: float32
- name: dhi
dtype: float32
splits:
- name: train
num_bytes: 450518979
num_examples: 290472
- name: test
num_bytes: 23313051
num_examples: 15288
download_size: 57111923
dataset_size: 473832030
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*
Dataset Card for Synth-Sky-Images
This dataset contains synthetic sky images generated using the Python package pvlib and some image manipulation. It is being used by me for testing and learning how to use HuggingFace's image dataset capabilities.
Dataset Details
Dataset Description
Images are generated by determining the azimuth and zenith angles at each timestamp and projecting onto a fisheye lens. North points vertically downward in the images. The irradiances are from a clear-sky model. The coordinates used to generate the sun angles and irradiances are (42.35909424627762, -71.09166390057037).