Tesseract doesn't care about whether things are metal or plastic or cotton, it tries to turn whatever image it is given into text via a pipeline of rules and shape classifiers for text. It succeeds when the text is distinct enough from the background so that it can do that job, and I suspect you will struggle a lot with this image unless you do some preprocessing.
I worked on a project using Tesseract for scanning airplane decals and we had a preprocessing pipeline consisting of gaussian blurring, adaptive thresholding and more.
You might look into background removal as a preprocessing stage for this kind of image to give Tesseract a better chance of seeing just the text:
Cheers