Summary of Do Generalised Classifiers Really Work on Human Drawn Sketches?, by Hmrishav Bandyopadhyay et al.
Do Generalised Classifiers really work on Human Drawn Sketches?
by Hmrishav Bandyopadhyay, Pinaki Nath Chowdhury, Aneeshan Sain, Subhadeep Koley, Tao Xiang, Ayan Kumar Bhunia, Yi-Zhe Song
First submitted to arxiv on: 4 Jul 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 combines large foundation models with human understanding of sketches to revolutionize generalized sketch representation learning, particularly in classification tasks. The approach generalizes across unknown categories (open-set) and traverses abstraction levels (good, bad, or doodles), addressing long-standing challenges in the sketch literature. To achieve this, the authors modify the CLIP model by introducing a novel auxiliary head for raster-to-vector sketch conversion, making it “sketch-aware.” They then fine-tune the model to capture different abstraction levels using a weighted codebook of prompt biases. This framework outperforms existing algorithms in both zero-shot and few-shot setups and across various abstraction boundaries. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper helps computers understand human sketches better by combining big AI models with what we know about drawing. It makes the computer “smarter” at recognizing and learning from different types of sketches, no matter how simple or complex they are. The authors use a special technique to teach the computer to recognize and categorize various sketch styles, making it more accurate and efficient in understanding and creating new sketches. |
Keywords
» Artificial intelligence » Classification » Few shot » Prompt » Representation learning » Zero shot
