Table 2 Performance comparison of the adaptive WebP, progressive-VQGAN, and progressive-hyperprior models on the Kodak dataset across various SNR values
SNR (dB) | Method | Throughput | PSNR | SSIM | Tavg | T99.9% |
|---|---|---|---|---|---|---|
| Â | Â | Mpps | dB | Â | ms | ms |
-10 | Adaptive Webp | 0.00 | – | – | – | – |
| Â | Progressive-VQGAN | 4.88 | 25.64 | 0.71 | 63.14 | 272.00 |
| Â | Progressive-Hyperprior | 18.00 | 24.99 | 0.68 | 21.52 | 108.00 |
-5 | Adaptive Webp | 0.00 | – | – | – | – |
| Â | Progressive-VQGAN | 34.13 | 25.83 | 0.72 | 11.14 | 94.00 |
| Â | Progressive-Hyperprior | 66.38 | 25.28 | 0.69 | 6.62 | 60.00 |
0 | Adaptive Webp | 74.63 | 27.47 | 0.75 | 5.37 | 265.00 |
| Â | Progressive-VQGAN | 101.25 | 26.20 | 0.73 | 4.31 | 34.00 |
| Â | Progressive-Hyperprior | 160.50 | 25.94 | 0.72 | 3.33 | 21.00 |
5 | Adaptive Webp | 454.50 | 27.63 | 0.76 | 1.82 | 71.00 |
| Â | Progressive-VQGAN | 205.13 | 26.31 | 0.73 | 2.67 | 16.00 |
| Â | Progressive-Hyperprior | 307.88 | 26.35 | 0.73 | 2.22 | 12.00 |
