• Corpus ID: 229188000

SAfE: Self-Attention Based Unsupervised Road Safety Classification in Hazardous Environments

  title={SAfE: Self-Attention Based Unsupervised Road Safety Classification in Hazardous Environments},
  author={Divya Kothandaraman and Rohan Chandra and Dinesh Manocha},
We present a novel approach SAfE that can identify parts of an outdoor scene that are safe for driving, based on attention models. Our formulation is designed for hazardous weather conditions that can impair the visibility of human drivers as well as autonomous vehicles, increasing the risk of accidents. Our approach is unsupervised and uses domain adaptation, with entropy minimization and attention transfer discriminators, to leverage the large amounts of labeled data corresponding to clear… 
NSS-VAEs: Generative Scene Decomposition for Visual Navigable Space Construction
This work proposes a new network, NSS-VAEs (Navigable Space Segmentation Variational AutoEncoders), a representation-learning-based framework to enable robots to learn the navigable space segmentation in an unsupervised manner and assesses the prediction uncertainty related to the unstructuredness of the scenes.
DensePASS: Dense Panoramic Semantic Segmentation via Unsupervised Domain Adaptation with Attention-Augmented Context Exchange
A generic framework for cross-domain panoramic semantic segmentation based on different variants of attention-augmented domain adaptation modules is built and improves the domain adaptation performance of two standard segmentation networks by 6.05% and 11.26% in Mean IoU.
Polyline Based Generative Navigable Space Segmentation for Autonomous Visual Navigation
Through extensive experiments, it is validated that the proposed Polyline Segmentation Variational AutoEncoder Networks (PSV-Nets) can learn the visual navigable space with high accuracy, even without any single label and can significantly outperform the state-of-the-art fully supervised learning based segmentation methods.


AdapNet: Adaptive semantic segmentation in adverse environmental conditions
This paper proposes a novel semantic segmentation architecture and the convoluted mixture of deep experts (CMoDE) fusion technique that enables a multi-stream deep neural network to learn features from complementary modalities and spectra, each of which are specialized in a subset of the input space.
No More Discrimination: Cross City Adaptation of Road Scene Segmenters
This work proposes an unsupervised learning approach to adapt road scene segmenters across different cities by advancing a joint global and class-specific domain adversarial learning framework, and shows that this method improves the performance of semantic segmentation in multiple cities across continents.
Fishyscapes: A Benchmark for Safe Semantic Segmentation in Autonomous Driving
Fishyscapes is presented, the first public benchmark for uncertainty estimation in the real-world task of semantic segmentation for urban driving and shows that anomaly detection is far from solved even for ordinary situations, while the benchmark allows measuring advancements beyond the state of the art.
SNE-RoadSeg: Incorporating Surface Normal Information into Semantic Segmentation for Accurate Freespace Detection
This paper introduces a novel module, named surface normal estimator (SNE), which can infer surface normal information from dense depth/disparity images with high accuracy and efficiency, and proposes a data-fusion CNN architecture, referred to as RoadSeg, which can extract and fuse features from both RGB images and the inferred surfacenormal information for accurate freespace detection.
City-Scale Road Audit System using Deep Learning
A system for city-scale road audit, using some of the most recent developments in deep learning and semantic segmentation, and a multi-step deep learning model that segments the road, subdivide the road further into defects, tags the frame for each defect and finally localizes the defects on a map gathered using GPS.
Segmentation of Drivable Road Using Deep Fully Convolutional Residual Network with Pyramid Pooling
A powerful 112-layer RPP model for monocular vision-based road detection based on the combination of fully convolutional network, residual learning, and pyramid pooling is proposed, which ranks second in both unmarked road and marked road tasks, fifth in multiple-marked-lane task, and third in combination task.
Semantic Foggy Scene Understanding with Synthetic Data
A complete pipeline to add synthetic fog to real, clear-weather images using incomplete depth information is developed and supervised learning with the authors' synthetic data significantly improves the performance of state-of-the-art CNN for SFSU on Foggy Driving.
Model Adaptation with Synthetic and Real Data for Semantic Dense Foggy Scene Understanding
A novel method is proposed, named Curriculum Model Adaptation (CMAda), which gradually adapts a semantic segmentation model from light synthetic fog to dense real fog in multiple steps, using both synthetic and real foggy data.
FCNs in the Wild: Pixel-level Adversarial and Constraint-based Adaptation
This paper introduces the first domain adaptive semantic segmentation method, proposing an unsupervised adversarial approach to pixel prediction problems, and outperforms baselines across different settings on multiple large-scale datasets.
Curriculum Model Adaptation with Synthetic and Real Data for Semantic Foggy Scene Understanding
A novel method, named Curriculum Model Adaptation (CMAda), which gradually adapts a semantic segmentation model from light synthetic fog to dense real fog in multiple steps, using both labeled synthetic Foggy data and unlabeled real foggy data.