Dual Attention Networks for Few-Shot Fine-Grained Recognition
Authors: Shu-Lin Xu, Faen Zhang, Xiu-Shen Wei, Jianhua Wang2911-2919
AAAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments on three popular fine-grained benchmark datasets show that our DUAL ATT-NET obviously outperforms other existing state-of-the-art methods. |
| Researcher Affiliation | Collaboration | Shu-Lin Xu1,2, Faen Zhang3, Xiu-Shen Wei1,2,4 , Jianhua Wang3 1School of Computer Science and Engineering, Nanjing University of Science and Technology 2State Key Laboratory of Integrated Services Networks, Xidian University 3AInnovation Technology Group Co., Ltd 4State Key Laboratory for Novel Software Technology, Nanjing University |
| Pseudocode | No | The paper describes its methodology in text and figures but does not include any pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not provide an explicit statement or link for open-source code availability for the described methodology. |
| Open Datasets | Yes | We conduct the experiments on three popular used benchmark datasets for few-shot fine-grained recognition, i.e., CUB Birds (Wah et al. 2011),Stanford Dogs (Khosla et al. 2011),Stanford Cars (Krause et al. 2013). |
| Dataset Splits | Yes | For each dataset, we follow (Wei et al. 2019; Huang et al. 2019, 2020) to randomly split its original image categories into two disjoint subsets: One as the auxiliary training set B, and the other as the FSFG testing set N, which is shown in Table 1. ... To mimic testing scenarios, all meta-training sets and testing sets contain CS = 5. Furthermore we follow (Zhu, Liu, and Jiang 2020) to set Ns = 1 (Ns = 5) for 1-shot recognition (5-shot recognition) and Nq is set to 15 in all settings. |
| Hardware Specification | Yes | We also gratefully acknowledge the support of Mind Spore, CANN (Compute Architecture for Neural Networks) and Ascend AI Processor used for this research. |
| Software Dependencies | No | The paper mentions 'Mind Spore' and 'CANN (Compute Architecture for Neural Networks)' but does not provide specific version numbers for any software dependencies. |
| Experiment Setup | Yes | For hyperparameters, we set δ = 0.6 in Eq. (4), and k = Pni u=1 Pni v=1 Au,v = n2 i t in Eq. (5). During meta-training, all of models are trained from scratch in an end-to-end manner. We use the Adam optimizer with initial learning rate of 0.001. The total number of episode is 200,000 and the learning rate is of reduced as 1/2 after each 50,000 episodes. We apply data augmentation, which includes random crops, random horizontal flips, and color jitter at the meta-training stage, as well as center crops at the testing stage, in all implemented experiments. |