heatmappy
Draw image and video heatmaps in python
Image
Video
Install
pip install heatmappy
Requirements
- matplotlib
- moviepy
- numpy
- Pillow
- PySide (optional: up to ~20% faster than Pillow alone)
Examples
Given some points (co-ordinates) and a base image
from heatmappy import Heatmapper
from PIL import Image
example_points = [(100, 20), (120, 25), (200, 50), (60, 300), (170, 250)]
example_img_path = 'cat.jpg'
example_img = Image.open(example_img_path)Draw a basic heatmap on the PIL image object
heatmapper = Heatmapper()
heatmap = heatmapper.heatmap_on_img(example_points, example_img)
heatmap.save('heatmap.png')
Draw a reveal heatmap, given the image path
heatmapper = Heatmapper(opacity=0.9, colours='reveal')
heatmap = heatmapper.heatmap_on_img_path(example_points, example_img_path)
heatmap.save('heatmap.png')
Draw a video heatmap
Input points are in the form (x, y, t) where t is in milliseconds.
example_vid = os.path.join('assets', 'some_video.mp4')
example_points = [(100, 100, 25), (112, 92, 67), (17, 100, 36)]
img_heatmapper = Heatmapper()
video_heatmapper = VideoHeatmapper(img_heatmapper)
heatmap_video = video_heatmapper.heatmap_on_video_path(
video_path=example_vid,
points=example_points
)
heatmap_video.write_videofile('out.mp4', bitrate="5000k", fps=24)Heatmap config
The following options are available (shown with their default values):
heatmapper = Heatmapper(
point_diameter=50, # the size of each point to be drawn
point_strength=0.2, # the strength, between 0 and 1, of each point to be drawn
opacity=0.65, # the opacity of the heatmap layer
colours='default', # 'default' or 'reveal'
# OR a matplotlib LinearSegmentedColorMap object
# OR the path to a horizontal scale image
grey_heatmapper='PIL' # The object responsible for drawing the points
# Pillow used by default, 'PySide' option available if installed
)
video_heatmapper = VideoHeatmapper(
heatmapper # the img heatmapper to use (like the heatmapper above, for example)
)Provided colour schemes
default
reveal
Coming soon
- Can specify different point size for each point plotted.
License
MIT License.