EFANet: Exchangeable Feature Alignment Network for Arbitrary Style Transfer

Authors: Zhijie Wu, Chunjin Song, Yang Zhou, Minglun Gong, Hui Huang12305-12312

AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Qualitative and quantitative experiments demonstrate the advantages of our approach.
Researcher Affiliation Academia Zhijie Wu,1 Chunjin Song,1 Yang Zhou,1 Minglun Gong,2 Hui Huang1 1Shenzhen University, 2University of Guelph
Pseudocode No The paper describes the architecture and processes but does not contain any pseudocode or algorithm blocks.
Open Source Code No The paper does not provide an explicit statement about releasing source code or a link to a code repository.
Open Datasets Yes Place365 database (Zhou et al. 2014) and Wi Ki Art dataset (Nichol 2016) are used for content and style images respectively, following (Sanakoyeu et1 al. 2018).
Dataset Splits No The paper describes the training data preparation ('randomly sampled patches of size 256 256') and mentions '100 test images' but does not specify a distinct validation split or explicit percentages for train/validation/test splits.
Hardware Specification Yes All results are obtained with a 12G Titan V GPU and averaged over 100 256 256 test images.
Software Dependencies No The paper states 'We implement our model with Tensorflow (Abadi et al. 2016)' but does not provide specific version numbers for TensorFlow or other software dependencies.
Experiment Setup Yes We choose Adam optimizer (Kingma and Ba 2014) with a batch size of 4 and a learning rate of 0.0001, and set the decay rates by default for 150000 iterations. and where the four weighting parameters are respectively set as 1, 7, 0.1 and 5 through out the experiments.