Since the human visual system is so highly sensitive to faces I think the best approach here would be to apply a facial detection algorithm, then boost the energy for the regions where faces are detected.

Basically you'd apply a heuristic that because faces are so special to the human visual system, the perceived energy of a face is higher than the pixels would otherwise indicate.

This would make seams avoid altering any faces in the scene.

The algorithm definitely has issues with faces.

One part I didn't get into is artificial weights. The original paper discusses "painting" certain regions of the image as high energy, driving seams away from those regions. This ties into faces when the paper suggests face detection to automatically paint faces with high energy, avoiding these issues in the first place!

The basic algorithm doesn't work well with faces, but you can use facial recognition to add a large penalty to carving through regions with faces. So it can be extended pretty easily to handle this.

Probably intentional, the seam carving algorithm has a tendency to make faces look distorted: http://www.cs.cmu.edu/afs/cs.cmu.edu/academic/class/15463-f1...

"Human faces? You can't handle human faces!"

https://www.youtube.com/watch?v=zqeIqJGdA58