Differentiable Auxiliary Learning for Sketch Re-Identification
Authors: Xingyu Liu, Xu Cheng, Haoyu Chen, Hao Yu, Guoying Zhao
AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments verify the superior performance of our DALNet over the state-of-the-art methods for Sketch Re-ID, and the generalization in sketch-based image retrieval and sketch-photo face recognition tasks. |
| Researcher Affiliation | Academia | 1School of Computer Science, Nanjing University of Information Science and Technology, China 2Center for Machine Vision and Signal Analysis, University of Oulu, Finland |
| Pseudocode | No | The paper does not contain any sections or figures explicitly labeled as "Pseudocode" or "Algorithm" with structured steps. |
| Open Source Code | No | The paper does not provide any explicit statement about releasing source code or a link to a code repository. |
| Open Datasets | Yes | The experiments are performed on five public sketch-based datasets: PKU-Sketch (Pang et al. 2018), QMUL-Shoe V2(Shoe V2) (Yu et al. 2021), QMUL-Chair V2 (Chair V2) (Yu et al. 2021), CUHK student (Tang and Wang 2002) and CUFSF (Zhang, Wang, and Tang 2011). ... According to the experimental setting in (Pang et al. 2018), 150 identities are utilized for training and the rest 50 identities for testing. ... The details of the training and testing sets can be found in (Yu et al. 2021). ... The CUHK student dataset has 188 face photo-sketch pairs, of which 88 pairs are selected for training and the rest 100 pairs for testing. For CUFSF dataset, consisting of 1194 face photo-sketch pairs, we randomly select 500 and 694 persons as train and test sets according to (Zhang, Wang, and Tang 2011), respectively. |
| Dataset Splits | Yes | According to the experimental setting in (Pang et al. 2018), 150 identities are utilized for training and the rest 50 identities for testing. ...The CUHK student dataset has 188 face photo-sketch pairs, of which 88 pairs are selected for training and the rest 100 pairs for testing. For CUFSF dataset, consisting of 1194 face photo-sketch pairs, we randomly select 500 and 694 persons as train and test sets according to (Zhang, Wang, and Tang 2011), respectively. |
| Hardware Specification | Yes | We implement our model on Pytorch framework with an NVIDIA RTX-3090 GPU. |
| Software Dependencies | No | The paper mentions "Pytorch framework" but does not specify a version number. It also does not list any other software dependencies with specific version numbers. |
| Experiment Setup | Yes | We employ the Adam W optimizer for 100 epochs with an initial learning rate of 0.00045 and a weight decay of 0.02. In the 3 3 convolutional kernel of our generator, the initial center weight is set to 9 and the surrounding weights are set to -0.8. The parameters ξ and λ are set to 0.07 and 0.6, respectively. The scale parameter γ is 64. The margin parameters mcm and mim are set to 0.25 and 0.5, respectively. |