Progressive Domain-Independent Feature Decomposition Network for Zero-Shot Sketch-Based Image Retrieval

Authors: Xinxun Xu, Muli Yang, Yanhua Yang, Hao Wang

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

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments demonstrate the superiority of our PDFD over state-of-the-art competitors. and Extensive experiments conducted on two popular large-scale datasets demonstrate that our proposed PDFD significantly outperforms state-of-the-art methods.
Researcher Affiliation Academia Xinxun Xu , Muli Yang , Yanhua Yang and Hao Wang Xidian University {xinxun.xu,muliyang.xd,haowang.xidian}@gmail.com, yanhyang@xidian.edu.cn
Pseudocode No No pseudocode or clearly labeled algorithm blocks were found.
Open Source Code No No explicit statement or link was found indicating the provision of open-source code for the described methodology.
Open Datasets Yes There are two widely-used large-scale sketch datasets Sketchy [Sangkloy et al., 2016] and TU-Berlin [Eitz et al., 2012] for ZS-SBIR. and VGG16 [Simonyan and Zisserman, 2014] pre-trained on Image Net [Deng et al., 2009] dataset (before the last pooling layer).
Dataset Splits No The paper defines a training set Dtr = {X seen, Sseen, Yseen} and a test set Dte = {X un, Yun} for the zero-shot setting, but does not explicitly provide percentages or specific counts for training, validation, and test splits for reproducibility. It states 'Following the same zero-shot data partitioning in SEMPCYC [Dutta and Akata, 2019]', but the exact numerical splits are not provided within this paper.
Hardware Specification No No specific hardware details (e.g., GPU/CPU models, memory amounts) used for running experiments were mentioned.
Software Dependencies No The paper mentions 'Py Torch' but does not provide a specific version number. Other mentioned software or models (e.g., Adam, VGG16, Word2Vec, GloVe) do not include version numbers for the specific implementation used.
Experiment Setup Yes PDFD is trained with Adam [Kingma and Ba, 2014] optimizer on Py Torch with an initial learning rate lr = 0.0001, β1 = 0.5, β2 = 0.99. The input size of the image/sketch is 224 224. We use the grid search method to select the best coefficients, which are λadv = 1.0, λrec = 1.0, λmcls = 1.0, λccls = 0.01 when training on Sketch and λadv = 1.0, λrec = 0.5, λmcls = 0.4, λccls = 0.4 when training on TUBerlin.