What Sketch Explainability Really Means for Downstream Tasks
Preprint
Abstract
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.Keywords
All Related Versions
This publication has 65 references indexed in Scilit:
- Bayesian safe policy learning with chance constrained optimization: application to military security assessment during the Vietnam WarJournal of the Royal Statistical Society Series A: Statistics in Society, 2025
- Evaluation of post-hoc interpretability methods in time-series classificationNature Machine Intelligence, 2023
- FS-COCO: Towards Understanding of Freehand Sketches of Common Objects in Context2022
- Garment Ideation: Iterative View-Aware Sketch-Based Garment Modeling2022
- Advances in Adversarial Attacks and Defenses in Computer Vision: A SurveyIEEE Access, 2021
- Semantic Hierarchy Emerges in Deep Generative Representations for Scene SynthesisInternational Journal of Computer Vision, 2021
- Stop explaining black box machine learning models for high stakes decisions and use interpretable models insteadNature Machine Intelligence, 2019
- SketchyScene: Richly-Annotated Scene Sketches2018
- Top-Down Neural Attention by Excitation Backprop2016
- SketchNet: Sketch Classification with Web Images2016