Help obtaining the position of the swatches in the original sample photo.

[vicgalba]

Hi, I’m working with this project to analyze different photo sessions. In these multiple photos there is a colorchecker on the same position, always. In the first photo of the session I run the segmentation/detection function with additional_data=True, so I can get the, principally, swatch colours.

Since the colour checker will always stay in the same place for the rest of the photos in the sessions, I’d like to avoid running the segmentation again on every image, but obtaining the swatches colours on every other photo with the data obtained from the first one. This, for doing a calibration of these next photos later.

With this context, what other attribute in the output you think could help me in my case? the documentation its great, but with all the contours and quadrilaterals obtained i got a little confuse, sorry.

Thanks in advance for the work done here, KelSolaar, its helping me a lot.

[KelSolaar]

Hi @vicgalba,

Check how its done here: https://github.com/ampas/idt-calculator/blob/master/aces/idt/generators/base_generator.py#L351C1-L365C2, the process is as follows:

• Reformat all the images at the working resolution
• Use additional=True to produce the rectangles
• Use the first rectangle along side with sample_colour_checker to extract the colours

[vicgalba]

Thank you, just what I needed.

I changed the format of the images, obtained the quadrilateral from the first image, and finally used this and the rectangle measurements in the documentation with sample_colour_checker on the rest of the photos in the session.

Can I ask you a slightly more general question?

I tried to build a calibration to transform photos from SRGB to Lab. To do this, I built a least squares regression to obtain a transformation equation, using the RGB values obtained from the color checker in each image and the reference Lab values.

My question is: how can I check that the values obtained make sense?

I looked at convert in the color library (https://colour.readthedocs.io/en/master/generated/colour.convert.html#colour.convert), but it doesn't mention anywhere a lab -> rgb conversion (to see if I can return to the original photo, confirming the validity of my method). Is this even possible?