I'm looking for a TensorFlow model detecting empty spaces on the images. I need to add my company logo to this empty area so there shouldn't be any faces or objects in this area. Also, I would be thankful for information if the is any existing solution for doing that (can be paid)
It sounds like you're looking for a foreground/background segmentation model (also called image matting). There are several ways to do this.
Here's one option for a Tensorflow model: https://github.com/adumrewal/imageMatting
Paperswithcode is a website compiling paper implementations along with their benchmark scores. You can find several tensorflow-based models on their Image Matting leaderboard.
However, if using Tensorflow isn't 100% necessary, then I'd go with something like PaddleSeg instead, which provides more well-tested implementations for applications like this, whereas running research repositories can often be a bit frustrating.