Multi-Graph Transformer for Free-Hand Sketch Recognition
This addresses the problem of sketch recognition for computer vision applications by introducing a novel graph-based approach, though it is incremental as it builds on existing GNN and transformer methods.
The paper tackles free-hand sketch recognition by representing sketches as multiple sparsely connected graphs and proposes the Multi-Graph Transformer (MGT) to learn from these graphs, achieving 72.80% accuracy on Google QuickDraw, which is close to the CNN-based upper bound of 74.22% and significantly outperforms RNN-based models.
Learning meaningful representations of free-hand sketches remains a challenging task given the signal sparsity and the high-level abstraction of sketches. Existing techniques have focused on exploiting either the static nature of sketches with Convolutional Neural Networks (CNNs) or the temporal sequential property with Recurrent Neural Networks (RNNs). In this work, we propose a new representation of sketches as multiple sparsely connected graphs. We design a novel Graph Neural Network (GNN), the Multi-Graph Transformer (MGT), for learning representations of sketches from multiple graphs which simultaneously capture global and local geometric stroke structures, as well as temporal information. We report extensive numerical experiments on a sketch recognition task to demonstrate the performance of the proposed approach. Particularly, MGT applied on 414k sketches from Google QuickDraw: (i) achieves small recognition gap to the CNN-based performance upper bound (72.80% vs. 74.22%), and (ii) outperforms all RNN-based models by a significant margin. To the best of our knowledge, this is the first work proposing to represent sketches as graphs and apply GNNs for sketch recognition. Code and trained models are available at https://github.com/PengBoXiangShang/multigraph_transformer.