Offer the benchmark and tools to evaluate AI-generated visual concepts including images, audios, 3d and videos.
This benchmark contains 13061 Asian facical images with 8 attributes: age, gender, hair, facial feature, angle of view, skin color, face block, and facial expression.
Answer: A fine-grained and targeted benchmark to illustrate the effectiveness of the generative model which can yield stylish individual pictures.
1、Our FFHQ-Attr-benchmark is developed based on the FFHQ-benchmark, thus thanks to https://github.com/NVlabs/ffhq-dataset.git.