Fast Interactive Object Annotation with Curve-GCN

Abstract

Manually labeling objects by tracing their boundaries is a laborious process. In Polygon-RNN++ the authors proposed Polygon-RNN that produces polygonal annotations in a recurrent manner using a CNN-RNN architecture, allowing interactive correction via humans-in-the-loop. We propose a new framework that alleviates the sequential nature of Polygon-RNN, by predicting all vertices simultaneously using a Graph Convolutional Network (GCN). Our model is trained end-to-end. It supports object annotation by either polygons or splines, facilitating labeling efficiency for both line-based and curved objects. We show that Curve-GCN outperforms all existing approaches in automatic mode, including the powerful PSP-DeepLab and is significantly more efficient in interactive mode than Polygon-RNN++. Our model runs at 29.3ms in automatic, and 2.6ms in interactive mode, making it 10x and 100x faster than Polygon-RNN++.

In Computer Vision and Pattern Recognition (CVPR), Long Beach, US, 2019

References in corpus (3)

• Geometric deep learning: going beyond Euclidean data
• DeepSpline: Data-Driven Reconstruction of Parametric Curves and Surfaces
• Active spline model: A shape based model-interactive segmentation

Cited by in corpus (13)

• Graph-Based Deep Learning for Medical Diagnosis and Analysis: Past, Present and Future
• Deep Marching Tetrahedra: a Hybrid Representation for High-Resolution 3D Shape Synthesis
• Learning Deformable Tetrahedral Meshes for 3D Reconstruction
• PolyTransform: Deep Polygon Transformer for Instance Segmentation
• Anatomy-Aware Siamese Network: Exploiting Semantic Asymmetry for Accurate Pelvic Fracture Detection in X-ray Images
• Topological based classification using graph convolutional networks
• DISIR: Deep Image Segmentation with Interactive Refinement
• Boundary Regularized Building Footprint Extraction From Satellite Images Using Deep Neural Network
• iCap: Interactive Image Captioning with Predictive Text
• ContourRender: Detecting Arbitrary Contour Shape For Instance Segmentation In One Pass
• DAGMapper: Learning to Map by Discovering Lane Topology
• Video Region Annotation with Sparse Bounding Boxes
• Learning to Predict More Accurate Text Instances for Scene Text Detection