AAKR: Adversarial Attack-based Knowledge Retention for Continual Semantic Segmentation
Authors: Zhidong Yu, Xiaoman Liu, Jiajun Hu, Zhenbo Shi, Wei Yang
AAAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Our comprehensive experiments demonstrate the efficacy of AAKR, and showcase that AAKR surpasses state-of-the-art competitors on benchmark datasets. ... Extensive experiments conducted on the benchmark dataset demonstrate that AAKR not only surpasses distillation-based approaches but also achieves superior results compared to replay-based methods, without a substantial increase in computational and storage requirements. ... Experiments Experimental Setup Datasets. We validate our method on benchmark datasets Pascal VOC2012 (Everingham et al. 2010) and ADE20k (Zhou et al. 2017). ... Quantitative Evaluation ... Comparisons of Training Time ... Ablation Study ... Qualitative Evaluation |
| Researcher Affiliation | Academia | 1School of Computer Science and Technology, University of Science and Technology of China, Hefei, China 2Hefei National Laboratory, University of Science and Technology of China, Hefei, China 3Suzhou Institute for Advanced Research, University of Science and Technology of China, Suzhou, China 4Laboratory for Advanced Computing and Intelligence Engineering, Wuxi, China EMAIL, EMAIL |
| Pseudocode | No | The paper describes the methodology using textual explanations and mathematical formulas, but it does not include any structured pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not contain any explicit statements about releasing source code or provide links to code repositories. |
| Open Datasets | Yes | Datasets. We validate our method on benchmark datasets Pascal VOC2012 (Everingham et al. 2010) and ADE20k (Zhou et al. 2017). |
| Dataset Splits | Yes | The Pascal VOC2012 dataset contains 20 object classes and one background. Its training and validation sets include 10,582 and 1,449 images, respectively. The ADE20k dataset contains 150 objects with 20,210 training images and 2,000 test images. ... For the Pascal VOC2012 dataset, we perform experiments in six settings, including adding 1 class after training 19 classes (19-1), adding 5 classes after training 15 classes (15-5), adding 5 classes sequentially after training 15 classes (15-1s), and more challenging settings of 10-10, 10-5s, and 10-1s. For the ADE20k dataset, we perform experiments in four settings, which are 100-50, 50-50s, 100-10s and 100-5. |
| Hardware Specification | No | The paper mentions using Deeplabv3 with ResNet-101 as the backbone but does not specify any hardware details like GPU models, CPU types, or memory. |
| Software Dependencies | No | We use Deeplabv3 (Chen et al. 2017) as the segmentation network with Res Net-101 (He et al. 2016) as the backbone, which is pre-trained on Image Net (Deng et al. 2009). The paper names specific software (Deeplabv3, ResNet-101, ImageNet) but does not provide version numbers for any of them. |
| Experiment Setup | Yes | The model is trained for 30 epochs on Pascal VOC2012 and 60 epochs on ADE20k, respectively. Moreover, λadv is 0.5. For the adversarial attack of added images, ϵ is 64, attack step k is 3, and α is ϵ/k. |