Semantic segmentation is one of the most fundamental problems in computer vision with significant impact on a wide variety of applications. Adversarial learning is shown to be an effective approach for improving semantic segmentation quality by enforcing higher-level pixel correlations and structural information. However, state-of-the-art semantic segmentation models cannot be easily plugged into an adversarial setting because they are not designed to accommodate convergence and stability issues in adversarial networks. We bridge this gap by building a conditional adversarial network with a state-of-the-art segmentation model (DeepLabv3+) at its core. To battle the stability issues, we introduce a novel lookahead adversarial learning (LoAd) approach with an embedded label map aggregation module. We focus on semantic segmentation models that run fast at inference for near real-time field applications. Through extensive experimentation, we demonstrate that the proposed solution can alleviate divergence issues in an adversarial semantic segmentation setting and results in considerable performance improvements (+5% in some classes) on the baseline for three standard datasets.

Traditional empirical risk minimization (ERM) for semantic segmentation can disproportionately advantage or disadvantage certain target classes in favor of an (unfair but) improved overall performance. Inspired by the recently introduced tilted ERM (TERM), we propose tilted cross-entropy (TCE) loss and adapt it to the semantic segmentation set-ting to minimize performance disparity among target classes and promote fairness. Through quantitative and qualitative performance analyses, we demonstrate that the proposed Stochastic TCE for semantic segmentation can offer improved overall fairness by efficiently minimizing the performance disparity among the target classes of Cityscapes.

In this work, we propose an interaction-driven approach streamlined to support and improve a wide range of real-time 2D interaction metaphors for arbitrarily large pointclouds based on detected primitive shapes. Rather than performing shape detection as a costly pre-processing step on the entire point cloud at once, a user-controlled interaction determines the region that is to be segmented next. By keeping the size of the region and the number of points small, the algorithm produces meaningful results and therefore feedback on the local geometry within a fraction of a second. We can apply these finding for improved picking and selection metaphors in large point clouds, and propose further novel shape-assisted.