Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

SketchGraphNet: A Memory-Efficient Hybrid Graph Transformer for Large-Scale Sketch Corpora Recognition

Shilong Chen, Mingyuan Li, Zhaoyang Wang, Zhonglin Ye, Haixing Zhao

TL;DR

This work proposes SketchGraphNet, a hybrid graph neural architecture that integrates local message passing with a memory-efficient global attention mechanism, without relying on auxiliary positional or structural encodings, to investigate large-scale sketch recognition from a graph-native perspective.

Abstract

This work investigates large-scale sketch recognition from a graph-native perspective, where free-hand sketches are directly modeled as structured graphs rather than raster images or stroke sequences. We propose SketchGraphNet, a hybrid graph neural architecture that integrates local message passing with a memory-efficient global attention mechanism, without relying on auxiliary positional or structural encodings. To support systematic evaluation, we construct SketchGraph, a large-scale benchmark comprising 3.44 million graph-structured sketches across 344 categories, with two variants (A and R) to reflect different noise conditions. Each sketch is represented as a spatiotemporal graph with normalized stroke-order attributes. On SketchGraph-A and SketchGraph-R, SketchGraphNet achieves Top-1 accuracies of 83.62% and 87.61%, respectively, under a unified training configuration. MemEffAttn further reduces peak GPU memory by over 40% and training time by more than 30% compared with Performer-based global attention, while maintaining comparable accuracy.

SketchGraphNet: A Memory-Efficient Hybrid Graph Transformer for Large-Scale Sketch Corpora Recognition

TL;DR

This work proposes SketchGraphNet, a hybrid graph neural architecture that integrates local message passing with a memory-efficient global attention mechanism, without relying on auxiliary positional or structural encodings, to investigate large-scale sketch recognition from a graph-native perspective.

Abstract

This work investigates large-scale sketch recognition from a graph-native perspective, where free-hand sketches are directly modeled as structured graphs rather than raster images or stroke sequences. We propose SketchGraphNet, a hybrid graph neural architecture that integrates local message passing with a memory-efficient global attention mechanism, without relying on auxiliary positional or structural encodings. To support systematic evaluation, we construct SketchGraph, a large-scale benchmark comprising 3.44 million graph-structured sketches across 344 categories, with two variants (A and R) to reflect different noise conditions. Each sketch is represented as a spatiotemporal graph with normalized stroke-order attributes. On SketchGraph-A and SketchGraph-R, SketchGraphNet achieves Top-1 accuracies of 83.62% and 87.61%, respectively, under a unified training configuration. MemEffAttn further reduces peak GPU memory by over 40% and training time by more than 30% compared with Performer-based global attention, while maintaining comparable accuracy.
Paper Structure (23 sections, 6 equations, 8 figures, 7 tables, 2 algorithms)

This paper contains 23 sections, 6 equations, 8 figures, 7 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Methods
  3. Notation and Overall Architecture
  4. Local--Global Hybrid Convolution Block
  5. Memory-Efficient Global Attention
  6. Computational Considerations
  7. Standard Self-Attention
  8. MemEffAttn with Non-negative Feature Mapping
  9. Implementation Details
  10. Dense Batching and Masking
  11. Blockwise Attention via xFormers
  12. Dataset Construction
  13. Results and Discussion
  14. Baseline Configurations and Comparative Framework
  15. Parameter Settings
  16. ...and 8 more sections

Figures (8)

  • Figure 1: Architecture of SketchGraphNet.
  • Figure 2: Architecture of MemEffAttn.
  • Figure 3: Statistical Comparison Between the A and R Versions of the SketchGraph Dataset: (a) Path-Length Distributions Across Categories; (b) Category-Level Density in the Space of Stroke Count Versus Path Length.
  • Figure 4: Structural Statistics Aggregated Across All Categories in the Dataset.
  • Figure 5: Mutually Exclusive Visualization of A$-$R and R Samples in the SketchGraph Dataset.
  • ...and 3 more figures