General repo for work with Berkeley Fire Lab https://pyro.turtlebasket.ml
 
 
 
 
 
 
Go to file
michael b7bb145089 WIP batch processing 2022-10-21 11:41:37 -07:00
examples add dennis image 2022-10-19 18:38:25 -07:00
images-input WIP batch processing 2022-10-21 11:41:37 -07:00
images-output WIP batch processing 2022-10-21 11:41:37 -07:00
matlab-version add matlab version 2022-10-12 12:14:13 -07:00
screenshots update example image 2022-10-20 17:40:18 -07:00
static add scaling factor & equation settings 2022-10-20 17:33:15 -07:00
templates add scaling factor & equation settings 2022-10-20 17:33:15 -07:00
.gitignore add scaling factor & equation settings 2022-10-20 17:33:15 -07:00
Makefile use pipenv in makefile 2022-10-12 12:27:37 -07:00
Pipfile WIP batch processing 2022-10-21 11:41:37 -07:00
Pipfile.lock WIP batch processing 2022-10-21 11:41:37 -07:00
README.md WIP batch processing 2022-10-21 11:41:37 -07:00
batch-process.py WIP batch processing 2022-10-21 11:41:37 -07:00
config.example.yaml WIP batch processing 2022-10-21 11:41:37 -07:00
flask_frontend.py add scaling factor & equation settings 2022-10-20 17:33:15 -07:00
ratio_pyrometry.py WIP batch processing 2022-10-21 11:41:37 -07:00

README.md

Pyrometry image processing

Interface Screenshots

Input View Results

Using the web version

  1. Go to pyro.turtlebasket.ml.
  2. Select an input image.
  3. Enter your DSLR camera settings.
  4. Click "Generate Heatmap".

Using the local (batch) version

Create a new config file:

cp config.example.yaml config.yaml

...then edit the values to match your DSLR camera settings. Standard config syntax is:

---
# camera settings
iso: 64
i-darkcurrent: 7.7
exposure-time: 0.5
f-stop: 2.4

# pyrometry settings
min-temp: 600
max-temp: 1200
scaling-factor: 0.55

# output settings
smoothing-radius: 2

Install dependencies:

pipenv install

...then load images into images-input. Run:

pipenv run python3 batch-process.py

...and find outputs in images-output.

Developing the web frontend

To serve in production:

gunicorn flask_frontend:app

To autoreload on source file changes:

gunicorn flask_frontend:app --reload

Temperature maps

Grayscale pyrometry: currently basic; uses grayscale opencv import, then just applies a jet filter. Doesn't yet copy the full impl in the paper.

Ratio pyrometry: pretty damn close to what's in the paper but it's very broken atm

Test image:

Ratio pyrometry result (with convolutional smoothing):

According to general researcher consensus, ratio pyrometry is supposed to be more accurate.

Grayscale pyrometry result: