Iterative Instance Segmentation

Abstract

Existing methods for pixel-wise labelling tasks generally disregard the underlying structure of labellings, often leading to predictions that are visually implausible. While incorporating structure into the model should improve prediction quality, doing so is challenging - manually specifying the form of structural constraints may be impractical and inference often becomes intractable even if structural constraints are given. We sidestep this problem by reducing structured prediction to a sequence of unconstrained prediction problems and demonstrate that this approach is capable of automatically discovering priors on shape, contiguity of region predictions and smoothness of region contours from data without any a priori specification. On the instance segmentation task, this method outperforms the state-of-the-art, achieving a mean ${\mathrm{AP}}^r$ of 63.6% at 50% overlap and 43.3% at 70% overlap.

13 pages, 10 figures; IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016

References in corpus (4)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Efficient Inference in Fully Connected CRFs with Gaussian Edge Potentials
• Conditional Random Fields as Recurrent Neural Networks
• Convolutional Feature Masking for Joint Object and Stuff Segmentation

Cited by in corpus (12)

• Beyond Skip Connections: Top-Down Modulation for Object Detection
• Annotating Object Instances with a Polygon-RNN
• Deep Watershed Transform for Instance Segmentation
• Deep Back-Projection Networks For Super-Resolution
• Object Detection Free Instance Segmentation With Labeling Transformations
• Pose2Instance: Harnessing Keypoints for Person Instance Segmentation
• Boundary-aware Instance Segmentation
• Efficient Coarse-to-Fine Non-Local Module for the Detection of Small Objects
• Error Correction for Dense Semantic Image Labeling
• SeGAN: Segmenting and Generating the Invisible
• Learning View Priors for Single-view 3D Reconstruction
• Detect, Replace, Refine: Deep Structured Prediction For Pixel Wise Labeling