Palette to match ADAS dataset?

[Emmanuel Participant]

Can someone from FLIR advise on what kind of palette/settings we should use to match the 8-bit thermal data from the ADAS dataset? We trained a model on the ADAS dataset, and want to retrain with data from FLIR One grayscale, 8-bit thermal images. What would the palette be, and can we set it using the already existing FLIR One app, or do we need to develop our own app to extract this 8-bit thermal image?

We noticed there are two classes in the mobile SDK, the first is PaletteManager/Palette and we want to know what the selection should be there to match the ADAS dataset when we are developing our own app. The second class is the ColorDistribution class, which we discovered has linear versus histogram equalization on the image and we are wondering what kind of pre-processing has been done to the ADAS data and if we need to select one of these, which option should we pick? Is there any other kind of pre-processing we will need to do to match the ADAS 8-bit thermal images?

Thank you,
Emmanuel

• This topic was modified 5 months, 2 weeks ago by  Emmanuel.
• This topic was modified 5 months, 2 weeks ago by  Emmanuel.

[Author]

Posts