Combinatorial Optimization for Panoptic Segmentation: A Fully Differentiable Approach

Abstract

We propose a fully differentiable architecture for simultaneous semantic and instance segmentation (a.k.a. panoptic segmentation) consisting of a convolutional neural network and an asymmetric multiway cut problem solver. The latter solves a combinatorial optimization problem that elegantly incorporates semantic and boundary predictions to produce a panoptic labeling. Our formulation allows to directly maximize a smooth surrogate of the panoptic quality metric by backpropagating the gradient through the optimization problem. Experimental evaluation shows improvement by backpropagating through the optimization problem w.r.t. comparable approaches on Cityscapes and COCO datasets. Overall, our approach shows the utility of using combinatorial optimization in tandem with deep learning in a challenging large scale real-world problem and showcases benefits and insights into training such an architecture.

arXiv commentTo be presented at NeurIPS 2021

References in corpus (15)

• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
• PyTorch: An Imperative Style, High-Performance Deep Learning Library
• Estimating or Propagating Gradients Through Stochastic Neurons for Conditional Computation
• Trainable Nonlinear Reaction Diffusion: A Flexible Framework for Fast and Effective Image Restoration
• Deep Equilibrium Models
• DeeperLab: Single-Shot Image Parser
• DetectoRS: Detecting Objects with Recursive Feature Pyramid and Switchable Atrous Convolution
• Learning Graphical Model Parameters with Approximate Marginal Inference
• Learning to Fuse Things and Stuff
• Deep Graph Matching via Blackbox Differentiation of Combinatorial Solvers
• Optimizing Rank-based Metrics with Blackbox Differentiation
• Learning with Differentiable Perturbed Optimizers
• Differentiable Perturb-and-Parse: Semi-Supervised Parsing with a Structured Variational Autoencoder
• Differentiable Top-k Operator with Optimal Transport
• Learning Randomly Perturbed Structured Predictors for Direct Loss Minimization