Disentangled Face Attribute Editing via Instance-Aware Latent Space Search

Authors: Yuxuan Han, Jiaolong Yang, Ying Fu

IJCAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results on both GAN-generated and real-world images collectively show that our method outperforms state-of-the-art methods proposed recently by a wide margin.
Researcher Affiliation Collaboration Yuxuan Han1 , Jiaolong Yang2 , Ying Fu1 1Beijing Institute of Technology 2Mircosoft Research Asia {hanyuxuan, fuying}@bit.edu.cn, jiaoyan@microsoft.com
Pseudocode No The paper does not contain any structured pseudocode or algorithm blocks.
Open Source Code Yes Code is available at https://github.com/yxuhan/IALS.
Open Datasets Yes We test our framework on the W space of Style GAN generator trained on the FFHQ [Karras et al., 2018] and Celeb A-HQ [Karras et al., 2019] datasets. The attribute classifiers H( ) are Res Net-18 [He et al., 2016] networks trained on the Celeb A dataset [Liu et al., 2015].
Dataset Splits No The paper mentions using FFHQ and CelebA-HQ for training the GAN generator and CelebA for attribute classifiers. It also describes sampling latent codes for evaluation and using DT metrics. However, it does not explicitly provide specific train/validation/test splits with percentages, counts, or references to predefined splits for its own experimental setup.
Hardware Specification No The paper does not provide any specific hardware details such as GPU models, CPU types, or memory used for running the experiments.
Software Dependencies No The paper mentions using ResNet-18 and ResNet-50 for attribute classifiers but does not specify any software dependencies with version numbers (e.g., Python, PyTorch, TensorFlow versions).
Experiment Setup Yes We empirically set the step size of incremental updating in our method to 0.1 in the following experiments. For each (λ1, λ2) pair, we set nmax = 20 (i.e. sample 20 points on the DT curve) and k = 0.1 and adopt the trapezoidal quadrature formula to approximate the integral in AUC computation defined in Eq. (7). In the following we simply use (λ1, λ2) = (0.75, 0) for our editing method.