Self-Attention-Based Edge Computing Model for Synthesis Image to Text through Next-Generation AI Mechanism

Table 1

Comparison of methods on the CUB dataset.

Method	IS	FID
GAN-INT-CLS [2]	2.88 ± 0.04	68.79
GAWWN [16]	3.62 ± 0.07	—
StackGAN [5]	3.70 ± 0.04	51.89
StackGAN-v2 [6]	3.84 ± 0.06	30.30
HDGAN [10]	4.15 ± 0.05	—
AttnGAN [7]	4.36 ± 0.03	15.38
Proposed approach	4.52 ± 0.03	14.25