AI Paper Atlas

Research map / 3D vision / NeRF and radiance fields

Learning 3D-aware GANs from Unposed Images with Template Feature Field

Xinya Chen, Hanlei Guo, Yanrui Bin, Shangzhan Zhang, Yuanbo Yang, Yujun Shen, Yue Wang, Yiyi Liao

ECCV 2024 · 3D vision › NeRF and radiance fields · approach: 3D-aware generative radiance fields · 1 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
estimate poses with a learned template feature field instead of letting the generator learn the pose distribution (vs 3DGP/PoF3D)3D-aware GAN (EG3D-based) · CompCarsDepth_360: 0.95 (EG3D), 4.02 (3DGP), 10.31 (PoF3D) -> 0.31Improved
use template feature field rather than template radiance field (RGB) for pose estimation (ablation)TeFF pose estimation · ShapeNet Carselevation KL divergence: 0.1422 -> 0.0696Improved
use template feature field rather than gray-image template radiance field (ablation)TeFF pose estimation · ShapeNet Carsazimuth KL divergence: 0.0656 -> 0.0555Improved
add phase correlation to model scale and in-plane rotation (4 DoF instead of 2 DoF) (ablation)TeFF · CompCarsDepth_360: 4.98 -> 0.31Improved
add phase correlation (4 DoF instead of 2 DoF) (ablation)TeFF · CompCarsFID_est: 11.09 -> 20.60Hurt
increase azimuth x elevation bins from 12x6 to 36x18 (ablation)TeFF · ShapeNet CarsFID_gt: 11.29 -> 5.95Improved
increase bins beyond 36x18 (to 60x30) (ablation)TeFF · ShapeNet CarsFID_gt: 5.95 -> 5.96No clear effect
use phase correlation instead of naive grid search over scale and rotation (ablation)TeFF · runtime (A6000)early-stage template rendering time (s/iter): 4423.0519 -> 0.0992Improved
estimate pose with TeFF versus EG3D with uniform pose samplingTeFF · CompCarsFID_360: 7.06 -> 27.71Hurt
TeFF pose estimation versus jointly learned pose distributionTeFF · ShapeNet Carsazimuth KL divergence: 4.4829 (PoF3D) -> 0.0555Improved
apply TeFF to face/cat data without GT posesTeFF · FFHQFID_gt: 4.80 (EG3D) -> 5.64Mixed

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy