An Improved Self-Training Model with Fine-Tuning Teacher/Student Exchange Strategy to Detect Computer-Generated Images

Figure 2

Original image sample and processed samples after different attacks. (a) No attack. (b) Noise. (c) Translation. (d) Scaling. (e) Partial content blocking. (f) Color channel change. (g) Affine transformation. (h) Blurring.

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)