Skip to main content

Flyr is a library for extracting thermal data from FLIR images written fully in Python, without depending on ExifTool.

Project description


A picture of a FLIR thermogram, the embedded optical data and a Flyr render concatenated into one

Flyr is a library for extracting thermal data from FLIR images written fully in Python.

Other solutions are wrappers around ExifTool to actually do the hard part of extracting the thermal data. Flyr is a reimplementation of the ExifTool's FLIR parser. Practically, this offers the following benefits:

  • Faster decoding because no new process needs to be started and in-memory data does not need to be communicated to this other process
  • More accurate, because Flyr uses all of the camera metadata to translate the raw values into Kelvin, while other projects have a certain set hardcoded. The differences are often about 0.1°C, but can be as high as 0.6°C.
  • Easier and robust installation and deployment, because Flyr is completely installable from PyPI.
  • Arguably simpler use: no need to create a superfluous extraction object; simply call thermogram = flyr.unpack(flir_file_path) and done
  • Extra features (see feature section) such as different units, built-in rendering and adjustable thermal data.


Flyr is installable from PyPi:

pip install flyr

Latest additional features

These features appeared in 3.1.0 through 3.3.0. Also see the for feature history.

  • Added ability to access some EXIF metadata via thermogram.camera_metadata (GPS, date, camera)
  • Added ability to mask renders or mask picture-in-pictures, thermally highlighting only the masked region
  • Added ability to render grayscale-inverted
  • Added ability to render picture-in-pictures
  • Added ability to emphasize edges in renders using the optical data
  • Added options to the flyr executable for the above two features

Upcoming features

These are features already implemented and merged into the master branch, but not released yet. Also see the for feature history.

Usage and features

Different units

Thermal data is available in kelvin, celsius and fahrenheit.

import flyr

flir_path = "thermograms/flir_c5_1.jpg"
thermogram = flyr.unpack(flir_path)

thermal = thermogram.kelvin  # As kelvin
thermal = thermogram.celsius  # As celsius
thermal = thermogram.fahrenheit  # As fahrenheit

Optical data can be read

The optical photo embedded in the FLIR thermogram

To read the embedded photo, access either optical or optical_pil to respectively get a 3D numpy or Pillow Image object with the photo.

import flyr

flir_path = "thermograms/flir_e5_2.jpg"
thermogram = flyr.unpack(flir_path)
optical_arr = thermogram.optical  # Also works"optical.jpg")

Built-in support for rendering

Examples of different RGB renders of the same thermogram

Flyr has built-in support to render thermal data to RGB images. It is possible to use the embedded palette or one of the provides palettes. Normalization can be done by percentiles or absolute values.

import flyr

flir_path = "thermograms/flir_e5_2.jpg"
thermogram = flyr.unpack(flir_path)
# render = thermogram.render()  # Use to get raw RGB array
render = thermogram.render_pil()  # Returns Pillow Image object"render-embedded.png")
palettes = ["turbo", "cividis", "inferno", "grayscale", "hot"]
for p in palettes:
    # The below call returns a Pillow Image object.
    # A sibling method called `render` returns a numpy array.
    render = thermogram.render_pil(

To render by percentiles, call as below. This approach is useful when it isn't known what temperature range to render.


It's also possible to apply a palette only to a specific part of rendered image, highlighting it and leaving the rest grayscale. Just pass a boolean mask to the mask parameter.

mask = thermogram.kelvin > thermogram.kelvin.mean()

Edge emphasis for better delineation

Example of five renders with and without edges emphasized

If optical imagery is present, it can be used to detect edges and more sharply delinate them. Used the edge_emphasis parameter with a value between 0 and 1 to enable it. When a mask is applied, edges are only emphasized in the masked region.

import flyr

thermogram = flyr.unpack("flir_e6_1.jpg")

mask = thermogram.kelvin > thermogram.kelvin.mean()
thermogram.render_pil(edge_emphasis=0.275, mask=mask).save("render-edge-emphasis-masked.png")

Putting the Picture in the Picture

Example of a render pictured in a photograph (Picture-in-Picture)

Renders and optical imagery can also be combined according to the Picture-in-Picture mode.

import flyr

thermogram = flyr.unpack("flir_e40_4.jpg")

Example of a masked render pictured in a photograph (Picture-in-Picture)

Like the render_pil method, a mask parameter can be used to highlight a certain region of the image. There are two modes in which this mask can work: "classical" and "alternative". In classical mode, the photograph is highlighted using a thermal mask of the region of interest. In the alternative mode, a thermal mask is applied to a grayscale render of the thermogram layered on top of the optical image.

import flyr

thermogram = flyr.unpack('flir_one_pro_1.jpg')
mask = thermogram.kelvin > thermogram.kelvin.mean()
thermogram.picture_in_picture_pil(mask=mask, mask_mode="classical").save("pip_classical.png")
thermogram.picture_in_picture_pil(mask=mask, mask_mode="alternative").save("pip_alternative.png")

Adjustable camera settings

Examples of different RGB renders of the same thermogram

It is possible to update the camera settings / parameters with which the thermal data is calculated. A typical value to adjust would be emissivity, but object_distance, relative_humidity and others are also configurable. See the parameters of FlyrThermogam.__raw_to_kelvin() for which.

import flyr

flir_path = "thermograms/flir_e5_2.jpg"
thermogram = flyr.unpack(flir_path)

emissivities = [0.6, 0.7, 0.8, 0.9, 1.0]
for e in emissivities:
    thermogram = thermogram.adjust_metadata(emissivity=e)
    # thermal = thermogram.celsius  # Access updated data as normal
    render = thermogram.render_pil(

Read from file, from file handle or binary stream

Call flyr.unpack on a filepath to receive a numpy array with the thermal data. Alternatively, first open the file in binary mode for reading and and pass the the file handle to flyr.unpack.

import flyr

# From file path
flir_path = "thermograms/flir_e5_2.jpg"
thermogram = flyr.unpack(flir_path)  # From file path
# From file handle / binary stream
with open(flir_path, "rb") as flir_handle:  # In binary mode!
    thermogram = flyr.unpack(flir_handle)

Access EXIF data

Some common EXIF metadata has been made easily accessible via thermogram.camera_metadata:

import flyr

# From file path
flir_path = "thermograms/flir_e75_1.jpg"
thermogram = flyr.unpack(flir_path)
cm = thermogram.camera_metadata
print(  # Raw EXIF data (dict)
print(cm.gps_data)  # Raw GPS data (dict)
print(cm.date_time)  # Parsed datetime object of when picture was taken (datetime)
print(cm.gps_altitude)  # (float)
print(cm.gps_image_direction)  # (float)
print(cm.gps_latitude)  # (float)
print(cm.gps_longitude)  # (float)
print(cm.gps_map_datum)  # (str)
print(cm.make)  # (str)
print(cm.model)  # (str)
print(  # (str)
print(cm.x_resolution)  # (float)
print(cm.y_resolution)  # (float)

The data and gps_data properties may also contain values not accessible via the handy properties mentioned above. The return values are always either the value itself or None, in case this EXIF data is not embedded in the file.

Supported cameras

Currently this library has been tested to work with:

  • FLIR C5
  • FLIR E4, E5, E6, E8, E8XT, E30, E30BX, E40, E50, E50BX, E53, E60BX, E75
  • FLIR I5
  • FLIR ONE, ONE Pro, ONE Pro Next Gen
  • FLIR P60 (PAL)
  • FLIR SC660
  • FLIR T630SC, T660
  • FLIR ThermaCAM B400


Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change. Most help is currently needed supporting more models and testing against more pictures. Testing and developing for your own camera's images or FLIR Tools' samples is recommended.


This code would not be possible without ExifTool's efforts to document the FLIR format. tomas123's work is similarly important to mention. Previous work in Python must also be acknowledged for creating a workable solution.


Flyr is licensed under The European Union Public License 1.2. The English version is included in the license file. Translations for all EU languages, each fully legally valid, can be found at the EUPL website.

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

flyr-3.3.1.tar.gz (48.0 kB view hashes)

Uploaded source

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page