sjtuplayer / few-shot-diffusion

[ICCV 2023] Phasic Content Fusing Diffusion Model with Directional Distribution Consistency for Few-Shot Model Adaption
50 stars 2 forks source link

code for Quantitative Evaluation #6

Open fikry102 opened 8 months ago

fikry102 commented 8 months ago

We first sample 1000 images from Style-GAN2 [15] as the source images and generate 1000 images in target domain by all the methods. Then we calculate the IS, IC-LPIPS and SCS on these generated images

Will the code for quantitative evaluation be released for more convenient comparison experiment with your method?

sjtuplayer commented 8 months ago

We first sample 1000 images from Style-GAN2 [15] as the source images and generate 1000 images in target domain by all the methods. Then we calculate the IS, IC-LPIPS and SCS on these generated images

Will the code for quantitative evaluation be released for more convenient comparison experiment with your method?

We use torch-fidelity to test the IS score, you can install it by running: pip install torch-fidelity Then you can calculate IS by: fidelity --gpu 0 --isc --input1 img_dir/ For IC-LPIPIS, you can refer to: https://github.com/utkarshojha/few-shot-gan-adaptation And for SCS, you can refer to: https://github.com/StevenShaw1999/RSSA