Summary of St-saclf: Style Transfer Informed Self-attention Classifier For Bias-aware Painting Classification, by Mridula Vijendran et al.
ST-SACLF: Style Transfer Informed Self-Attention Classifier for Bias-Aware Painting Classification
by Mridula Vijendran, Frederick W. B. Li, Jingjing Deng, Hubert P. H. Shum
First submitted to arxiv on: 3 Aug 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper tackles the challenge of painting classification in digital and classic art galleries. Existing methods struggle with adapting real-world knowledge to artistic images during training, leading to poor performance when dealing with different datasets. The proposed two-step process addresses this issue by generating more data using Style Transfer with Adaptive Instance Normalization (AdaIN) and then improving the classifier’s understanding of artistic details through feature-map adaptive spatial attention modules. Additionally, the paper tackles imbalanced class representation by dynamically adjusting augmented samples. The study achieves an impressive 87.24% accuracy using a ResNet-50 backbone over 40 training epochs and explores quantitative analyses comparing different pretrained backbones, model optimization through ablation studies, and varying augmentation levels affecting model performance. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This research helps art galleries organize and find artwork more effectively. Right now, computers struggle to understand the difference between artistic images from various styles and eras. The scientists came up with a two-step solution. First, they created new training data that combines different art styles. Then, they developed an improved computer model that can focus on important details in artwork. This makes it better at recognizing paintings. The study also looks at how to make the model work well with different datasets and explores what happens when it’s trained with more or less information. |
Keywords
» Artificial intelligence » Attention » Classification » Feature map » Optimization » Resnet » Style transfer
