What Sketch Explainability Really Means for Downstream Tasks
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866911797993275392 |
|---|---|
| author | Bandyopadhyay, Hmrishav Chowdhury, Pinaki Nath Bhunia, Ayan Kumar Sain, Aneeshan Xiang, Tao Song, Yi-Zhe |
| author_facet | Bandyopadhyay, Hmrishav Chowdhury, Pinaki Nath Bhunia, Ayan Kumar Sain, Aneeshan Xiang, Tao Song, Yi-Zhe |
| contents | In this paper, we explore the unique modality of sketch for explainability, emphasising the profound impact of human strokes compared to conventional pixel-oriented studies. Beyond explanations of network behavior, we discern the genuine implications of explainability across diverse downstream sketch-related tasks. We propose a lightweight and portable explainability solution -- a seamless plugin that integrates effortlessly with any pre-trained model, eliminating the need for re-training. Demonstrating its adaptability, we present four applications: highly studied retrieval and generation, and completely novel assisted drawing and sketch adversarial attacks. The centrepiece to our solution is a stroke-level attribution map that takes different forms when linked with downstream tasks. By addressing the inherent non-differentiability of rasterisation, we enable explanations at both coarse stroke level (SLA) and partial stroke level (P-SLA), each with its advantages for specific downstream tasks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2403_09480 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | What Sketch Explainability Really Means for Downstream Tasks Bandyopadhyay, Hmrishav Chowdhury, Pinaki Nath Bhunia, Ayan Kumar Sain, Aneeshan Xiang, Tao Song, Yi-Zhe Computer Vision and Pattern Recognition Artificial Intelligence In this paper, we explore the unique modality of sketch for explainability, emphasising the profound impact of human strokes compared to conventional pixel-oriented studies. Beyond explanations of network behavior, we discern the genuine implications of explainability across diverse downstream sketch-related tasks. We propose a lightweight and portable explainability solution -- a seamless plugin that integrates effortlessly with any pre-trained model, eliminating the need for re-training. Demonstrating its adaptability, we present four applications: highly studied retrieval and generation, and completely novel assisted drawing and sketch adversarial attacks. The centrepiece to our solution is a stroke-level attribution map that takes different forms when linked with downstream tasks. By addressing the inherent non-differentiability of rasterisation, we enable explanations at both coarse stroke level (SLA) and partial stroke level (P-SLA), each with its advantages for specific downstream tasks. |
| title | What Sketch Explainability Really Means for Downstream Tasks |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence |
| url | https://arxiv.org/abs/2403.09480 |