The Cityscapes Dataset for Semantic Urban Scene Understanding

Abstract

Visual understanding of complex urban street scenes is an enabling factor for a wide range of applications. Object detection has benefited enormously from large-scale datasets, especially in the context of deep learning. For semantic urban scene understanding, however, no current dataset adequately captures the complexity of real-world urban scenes. To address this, we introduce Cityscapes, a benchmark suite and large-scale dataset to train and test approaches for pixel-level and instance-level semantic labeling. Cityscapes is comprised of a large, diverse set of stereo video sequences recorded in streets from 50 different cities. 5000 of these images have high quality pixel-level annotations; 20000 additional images have coarse annotations to enable methods that leverage large volumes of weakly-labeled data. Crucially, our effort exceeds previous attempts in terms of dataset size, annotation richness, scene variability, and complexity. Our accompanying empirical study provides an in-depth analysis of the dataset characteristics, as well as a performance evaluation of several state-of-the-art approaches based on our benchmark.

arXiv commentIncludes supplemental material

References in corpus (6)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• ParseNet: Looking Wider to See Better
• STC: A Simple to Complex Framework for Weakly-supervised Semantic Segmentation
• Fully Connected Deep Structured Networks
• What's the Point: Semantic Segmentation with Point Supervision
• Semantic Instance Annotation of Street Scenes by 3D to 2D Label Transfer

Cited by in corpus (435)

• Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation
• A Review on Deep Learning Techniques Applied to Semantic Segmentation
• ENet: A Deep Neural Network Architecture for Real-Time Semantic Segmentation
• Toward Multimodal Image-to-Image Translation
• Momentum Contrast for Unsupervised Visual Representation Learning
• Learning to Reweight Examples for Robust Deep Learning
• FCNs in the Wild: Pixel-level Adversarial and Constraint-based Adaptation
• CyCADA: Cycle-Consistent Adversarial Domain Adaptation
• Automatic lung segmentation in routine imaging is primarily a data diversity problem, not a methodology problem
• OCNet: Object Context Network for Scene Parsing
• Multi-Task Learning Using Uncertainty to Weigh Losses for Scene Geometry and Semantics
• Semantic Instance Segmentation with a Discriminative Loss Function
• Deformable Convolutional Networks
• Adversarial Learning for Semi-Supervised Semantic Segmentation
• Fast-SCNN: Fast Semantic Segmentation Network
• Evolution of Image Segmentation using Deep Convolutional Neural Network: A Survey
• Coordinate Attention for Efficient Mobile Network Design
• Unsupervised CNN for Single View Depth Estimation: Geometry to the Rescue
• Pyramid Scene Parsing Network
• Multimodal Unsupervised Image-to-Image Translation
• Cross-view Semantic Segmentation for Sensing Surroundings
• Tent: Fully Test-time Adaptation by Entropy Minimization
• Fully Convolutional Networks for Dense Semantic Labelling of High-Resolution Aerial Imagery
• Pyramid Attention Network for Semantic Segmentation
• Unsupervised Learning of Depth and Ego-Motion from Video
• AADS: Augmented Autonomous Driving Simulation using Data-driven Algorithms
• Assessing bikeability with street view imagery and computer vision
• ContextNet: Exploring Context and Detail for Semantic Segmentation in Real-time
• K-Net: Towards Unified Image Segmentation
• Semantic Understanding of Scenes through the ADE20K Dataset
• RedNet: Residual Encoder-Decoder Network for indoor RGB-D Semantic Segmentation
• Places: An Image Database for Deep Scene Understanding
• Dilated Residual Networks
• Differentiable Learning-to-Normalize via Switchable Normalization
• Playing for Data: Ground Truth from Computer Games
• Polarized Self-Attention: Towards High-quality Pixel-wise Regression
• FusionSeg: Learning to combine motion and appearance for fully automatic segmention of generic objects in videos
• The Fishyscapes Benchmark: Measuring Blind Spots in Semantic Segmentation
• Virtual to Real Reinforcement Learning for Autonomous Driving
• Maximum Classifier Discrepancy for Unsupervised Domain Adaptation
• ImageNet pre-trained models with batch normalization
• Masked-attention Mask Transformer for Universal Image Segmentation
• Benchmarking Robustness in Object Detection: Autonomous Driving when Winter is Coming
• ICNet for Real-Time Semantic Segmentation on High-Resolution Images
• HRFormer: High-Resolution Transformer for Dense Prediction
• SPG-Net: Segmentation Prediction and Guidance Network for Image Inpainting
• BiSeNet: Bilateral Segmentation Network for Real-time Semantic Segmentation
• Video-to-Video Synthesis
• GeoNet: Unsupervised Learning of Dense Depth, Optical Flow and Camera Pose
• Learning to Generate Images of Outdoor Scenes from Attributes and Semantic Layouts
• Light-Weight RefineNet for Real-Time Semantic Segmentation
• Tracking the Trackers: An Analysis of the State of the Art in Multiple Object Tracking
• One-Shot Unsupervised Cross Domain Translation
• Learning Linear Transformations for Fast Arbitrary Style Transfer
• Adversarial Examples for Semantic Segmentation and Object Detection
• Semi-Supervised Deep Learning for Monocular Depth Map Prediction
• Learning a Discriminative Feature Network for Semantic Segmentation
• CGNet: A Light-weight Context Guided Network for Semantic Segmentation
• Photographic Image Synthesis with Cascaded Refinement Networks
• DSP: Dual Soft-Paste for Unsupervised Domain Adaptive Semantic Segmentation
• FieldSAFE: Dataset for Obstacle Detection in Agriculture
• Depth Adaptive Deep Neural Network for Semantic Segmentation
• Domain Stylization: A Strong, Simple Baseline for Synthetic to Real Image Domain Adaptation
• Illuminating Pedestrians via Simultaneous Detection & Segmentation
• Semantic-aware Grad-GAN for Virtual-to-Real Urban Scene Adaption
• CityPersons: A Diverse Dataset for Pedestrian Detection
• Adversarial Style Mining for One-Shot Unsupervised Domain Adaptation
• Promises and pitfalls of deep neural networks in neuroimaging-based psychiatric research
• RefineNet: Multi-Path Refinement Networks for High-Resolution Semantic Segmentation
• PAD-Net: Multi-Tasks Guided Prediction-and-Distillation Network for Simultaneous Depth Estimation and Scene Parsing
• Recurrent Neural Networks for Semantic Instance Segmentation
• Selective Feature Connection Mechanism: Concatenating Multi-layer CNN Features with a Feature Selector
• Fully Convolutional Adaptation Networks for Semantic Segmentation
• OFFSEG: A Semantic Segmentation Framework For Off-Road Driving
• Adversarial Examples for Semantic Image Segmentation
• Driving Scene Perception Network: Real-time Joint Detection, Depth Estimation and Semantic Segmentation
• Real-time Kinematic Ground Truth for the Oxford RobotCar Dataset
• Small-scale Pedestrian Detection Based on Somatic Topology Localization and Temporal Feature Aggregation
• Adversarial Dropout Regularization
• A Commute in Data: The comma2k19 Dataset
• Syn2Real: A New Benchmark forSynthetic-to-Real Visual Domain Adaptation
• ShuffleSeg: Real-time Semantic Segmentation Network
• Global Aggregation then Local Distribution in Fully Convolutional Networks
• Stereo DSO: Large-Scale Direct Sparse Visual Odometry with Stereo Cameras
• Real-Time Semantic Segmentation via Multiply Spatial Fusion Network
• Strip Pooling: Rethinking Spatial Pooling for Scene Parsing
• Annotating Object Instances with a Polygon-RNN
• Domain Adaptation for Semantic Segmentation with Maximum Squares Loss
• The Lovász-Softmax loss: A tractable surrogate for the optimization of the intersection-over-union measure in neural networks
• End-to-end Learning of Driving Models from Large-scale Video Datasets
• Procedural Modeling and Physically Based Rendering for Synthetic Data Generation in Automotive Applications
• Relation-Aware Global Attention for Person Re-identification
• COCO-Stuff: Thing and Stuff Classes in Context
• Few-Shot Object Detection with Attention-RPN and Multi-Relation Detector
• Exemplar Guided Unsupervised Image-to-Image Translation with Semantic Consistency
• NASNet: A Neuron Attention Stage-by-Stage Net for Single Image Deraining
• Deep Watershed Transform for Instance Segmentation
• Weakly-Supervised Semantic Segmentation by Iteratively Mining Common Object Features
• Monocular 3D Object Detection with Pseudo-LiDAR Point Cloud
• Pixel-wise Attentional Gating for Parsimonious Pixel Labeling
• Predicting Deeper into the Future of Semantic Segmentation
• Where are the Masks: Instance Segmentation with Image-level Supervision
• Cityscapes 3D: Dataset and Benchmark for 9 DoF Vehicle Detection
• Learning Monocular Depth by Distilling Cross-domain Stereo Networks
• Stacked U-Nets: A No-Frills Approach to Natural Image Segmentation
• Laplacian Pyramid Reconstruction and Refinement for Semantic Segmentation
• Real-time Joint Object Detection and Semantic Segmentation Network for Automated Driving
• Diversify and Match: A Domain Adaptive Representation Learning Paradigm for Object Detection
• Context-self contrastive pretraining for crop type semantic segmentation
• Future Semantic Segmentation with Convolutional LSTM
• Exploring Object Relation in Mean Teacher for Cross-Domain Detection
• Natural Environment Benchmarks for Reinforcement Learning
• CCVS: Context-aware Controllable Video Synthesis
• Drop to Adapt: Learning Discriminative Features for Unsupervised Domain Adaptation
• Budgeted Training: Rethinking Deep Neural Network Training Under Resource Constraints
• On the Benefit of Adversarial Training for Monocular Depth Estimation
• Instance Segmentation by Deep Coloring
• Non-Homogeneous Haze Removal via Artificial Scene Prior and Bidimensional Graph Reasoning
• Decoupled Adaptation for Cross-Domain Object Detection
• Boundary IoU: Improving Object-Centric Image Segmentation Evaluation
• Text as Neural Operator: Image Manipulation by Text Instruction
• Interpretability Beyond Classification Output: Semantic Bottleneck Networks
• The highD Dataset: A Drone Dataset of Naturalistic Vehicle Trajectories on German Highways for Validation of Highly Automated Driving Systems
• InstaBoost: Boosting Instance Segmentation via Probability Map Guided Copy-Pasting
• Chainer: A Deep Learning Framework for Accelerating the Research Cycle
• Iteratively Trained Interactive Segmentation
• AI Benchmark: Running Deep Neural Networks on Android Smartphones
• Autonomous Driving in Reality with Reinforcement Learning and Image Translation
• Camera-based vehicle velocity estimation from monocular video
• DAiSEE: Towards User Engagement Recognition in the Wild
• Image to Image Translation for Domain Adaptation
• CEREALS - Cost-Effective REgion-based Active Learning for Semantic Segmentation
• Not All Pixels Are Equal: Difficulty-aware Semantic Segmentation via Deep Layer Cascade
• Significance-aware Information Bottleneck for Domain Adaptive Semantic Segmentation
• HR-Depth: High Resolution Self-Supervised Monocular Depth Estimation
• Reinforced active learning for image segmentation
• JointGAN: Multi-Domain Joint Distribution Learning with Generative Adversarial Nets
• CASENet: Deep Category-Aware Semantic Edge Detection
• No More Discrimination: Cross City Adaptation of Road Scene Segmenters
• InstanceCut: from Edges to Instances with MultiCut
• Semantic Instance Annotation of Street Scenes by 3D to 2D Label Transfer
• Adapting Semantic Segmentation Models for Changes in Illumination and Camera Perspective
• VRUNet: Multi-Task Learning Model for Intent Prediction of Vulnerable Road Users
• Self-Supervised Joint Learning Framework of Depth Estimation via Implicit Cues
• Reducing Information Bottleneck for Weakly Supervised Semantic Segmentation
• End-to-End Multi-Task Learning with Attention
• Controllable Pareto Multi-Task Learning
• Efficient Interactive Annotation of Segmentation Datasets with Polygon-RNN++
• LEGO: Learning Edge with Geometry all at Once by Watching Videos
• A Novel Upsampling and Context Convolution for Image Semantic Segmentation
• Repulsion Loss: Detecting Pedestrians in a Crowd
• Delving Deeper into Anti-aliasing in ConvNets
• Wisdom of Committees: An Overlooked Approach To Faster and More Accurate Models
• DROGON: A Trajectory Prediction Model based on Intention-Conditioned Behavior Reasoning
• Knowledge Adaptation for Efficient Semantic Segmentation
• Unsupervised Learning of Monocular Depth Estimation with Bundle Adjustment, Super-Resolution and Clip Loss
• Close Yet Distinctive Domain Adaptation
• Effective Use of Dilated Convolutions for Segmenting Small Object Instances in Remote Sensing Imagery
• All about Structure: Adapting Structural Information across Domains for Boosting Semantic Segmentation
• TorontoCity: Seeing the World with a Million Eyes
• Effective Use of Synthetic Data for Urban Scene Semantic Segmentation
• Towards Realistic 3D Embedding via View Alignment
• Geometry-Consistent Generative Adversarial Networks for One-Sided Unsupervised Domain Mapping
• Efficient semantic image segmentation with superpixel pooling
• Visibility Guided NMS: Efficient Boosting of Amodal Object Detection in Crowded Traffic Scenes
• Boundary-Aware Network for Fast and High-Accuracy Portrait Segmentation
• Can we unify monocular detectors for autonomous driving by using the pixel-wise semantic segmentation of CNNs?
• Rethinking Co-design of Neural Architectures and Hardware Accelerators
• Unified Perceptual Parsing for Scene Understanding
• Full-Resolution Residual Networks for Semantic Segmentation in Street Scenes
• Sparse and Dense Data with CNNs: Depth Completion and Semantic Segmentation
• Graph-Based Global Reasoning Networks
• Crowd Counting via Weighted VLAD on Dense Attribute Feature Maps
• Adaptive Semantic Segmentation with a Strategic Curriculum of Proxy Labels
• Semantic Single-Image Dehazing
• Rethinking Spatially-Adaptive Normalization
• Pre-Trained Models: Past, Present and Future
• CSPN++: Learning Context and Resource Aware Convolutional Spatial Propagation Networks for Depth Completion
• BLVD: Building A Large-scale 5D Semantics Benchmark for Autonomous Driving
• Scene Parsing with Global Context Embedding
• Model-driven Simulations for Deep Convolutional Neural Networks
• Joint Graph Decomposition and Node Labeling: Problem, Algorithms, Applications
• CaTDet: Cascaded Tracked Detector for Efficient Object Detection from Video
• Single View Stereo Matching
• AdaStereo: A Simple and Efficient Approach for Adaptive Stereo Matching
• Hybrid Learning of Optical Flow and Next Frame Prediction to Boost Optical Flow in the Wild
• Multi-Target Domain Adaptation with Collaborative Consistency Learning
• Learning to Steer by Mimicking Features from Heterogeneous Auxiliary Networks
• Semantic Video Segmentation by Gated Recurrent Flow Propagation
• Adaptive Affinity Fields for Semantic Segmentation
• Road Network and Travel Time Extraction from Multiple Look Angles with SpaceNet Data
• Exploring Context with Deep Structured models for Semantic Segmentation
• Sequential Adversarial Learning for Self-Supervised Deep Visual Odometry
• PatchFCN for Intracranial Hemorrhage Detection
• Discriminative Region Proposal Adversarial Networks for High-Quality Image-to-Image Translation
• Fully Convolutional Networks for Panoptic Segmentation
• Unbiasing Semantic Segmentation For Robot Perception using Synthetic Data Feature Transfer
• Learning Deep Representations for Semantic Image Parsing: a Comprehensive Overview
• Fused DNN: A deep neural network fusion approach to fast and robust pedestrian detection
• Understanding Visual Ads by Aligning Symbols and Objects using Co-Attention
• Customizable Architecture Search for Semantic Segmentation
• Semi-parametric Image Synthesis
• Semantic Correlation Promoted Shape-Variant Context for Segmentation
• Two Stream Networks for Self-Supervised Ego-Motion Estimation
• Temporally Distributed Networks for Fast Video Semantic Segmentation
• SegSort: Segmentation by Discriminative Sorting of Segments
• Student Becoming the Master: Knowledge Amalgamation for Joint Scene Parsing, Depth Estimation, and More
• PIRM Challenge on Perceptual Image Enhancement on Smartphones: Report
• X-Distill: Improving Self-Supervised Monocular Depth via Cross-Task Distillation
• Disentangling Propagation and Generation for Video Prediction
• Answering Visual What-If Questions: From Actions to Predicted Scene Descriptions
• Segmenting Transparent Objects in the Wild
• T2Net: Synthetic-to-Realistic Translation for Solving Single-Image Depth Estimation Tasks
• FoveaNet: Perspective-aware Urban Scene Parsing
• Recurrent Neural Network for Learning DenseDepth and Ego-Motion from Video
• Challenges in Monocular Visual Odometry: Photometric Calibration, Motion Bias and Rolling Shutter Effect
• A Review of Visual Odometry Methods and Its Applications for Autonomous Driving
• Tree-structured Kronecker Convolutional Network for Semantic Segmentation
• Boundary-aware Instance Segmentation
• Deep Feature Flow for Video Recognition
• Find Your Own Way: Weakly-Supervised Segmentation of Path Proposals for Urban Autonomy
• Semi-supervised Domain Adaptation based on Dual-level Domain Mixing for Semantic Segmentation
• LIMO: Lidar-Monocular Visual Odometry
• Paying More Attention to Saliency: Image Captioning with Saliency and Context Attention
• Revisiting Hierarchical Approach for Persistent Long-Term Video Prediction
• Scene Gated Social Graph: Pedestrian Trajectory Prediction Based on Dynamic Social Graphs and Scene Constraints
• Branched Multi-Task Networks: Deciding What Layers To Share
• PointFlow: Flowing Semantics Through Points for Aerial Image Segmentation
• WaveSNet: Wavelet Integrated Deep Networks for Image Segmentation
• Deep Virtual Stereo Odometry: Leveraging Deep Depth Prediction for Monocular Direct Sparse Odometry
• Few-shot Video-to-Video Synthesis
• BoundarySqueeze: Image Segmentation as Boundary Squeezing
• Training Bit Fully Convolutional Network for Fast Semantic Segmentation
• Multi-Scale Feature Aggregation by Cross-Scale Pixel-to-Region Relation Operation for Semantic Segmentation
• Unsupervised Domain Adaptation for Semantic Segmentation via Low-level Edge Information Transfer
• Adaptation Across Extreme Variations using Unlabeled Domain Bridges
• Learning 3D Human Pose from Structure and Motion
• Deep Convolutional Neural Networks with Spatial Regularization, Volume and Star-shape Priori for Image Segmentation
• Monocular Depth Estimators: Vulnerabilities and Attacks
• Workshop on Autonomous Driving at CVPR 2021: Technical Report for Streaming Perception Challenge
• Object category learning and retrieval with weak supervision
• Surveillance Video Parsing with Single Frame Supervision
• On The State of Data In Computer Vision: Human Annotations Remain Indispensable for Developing Deep Learning Models
• Deep Active Learning for Efficient Training of a LiDAR 3D Object Detector
• Experimental Force-Torque Dataset for Robot Learning of Multi-Shape Insertion
• Visuomotor Understanding for Representation Learning of Driving Scenes
• TransFlow: Unsupervised Motion Flow by Joint Geometric and Pixel-level Estimation
• Densely connected multidilated convolutional networks for dense prediction tasks
• Semantic Road Layout Understanding by Generative Adversarial Inpainting
• Robust Semantic Segmentation with Superpixel-Mix
• A Brief Review of Deep Multi-task Learning and Auxiliary Task Learning
• Learning to Navigate Autonomously in Outdoor Environments : MAVNet
• Visual-Inertial-Semantic Scene Representation for 3-D Object Detection
• Mix-and-Match Tuning for Self-Supervised Semantic Segmentation
• Exploring Multi-Branch and High-Level Semantic Networks for Improving Pedestrian Detection
• Obstacle Detection Quality as a Problem-Oriented Approach to Stereo Vision Algorithms Estimation in Road Situation Analysis
• Do Normalization Layers in a Deep ConvNet Really Need to Be Distinct?
• Efficacy of Pixel-Level OOD Detection for Semantic Segmentation
• Semantically-Aware Attentive Neural Embeddings for Image-based Visual Localization
• Mask-based Data Augmentation for Semi-supervised Semantic Segmentation
• Real-time Semantic Segmentation with Fast Attention
• DeLS-3D: Deep Localization and Segmentation with a 3D Semantic Map
• Dynamic Neural Representational Decoders for High-Resolution Semantic Segmentation
• SegStereo: Exploiting Semantic Information for Disparity Estimation
• Modular Vehicle Control for Transferring Semantic Information Between Weather Conditions Using GANs
• Can Ground Truth Label Propagation from Video help Semantic Segmentation?
• On the Diversity of Realistic Image Synthesis
• Distantly Supervised Road Segmentation
• Irregular Convolutional Neural Networks
• BusyHands: A Hand-Tool Interaction Database for Assembly Tasks Semantic Segmentation
• Coarse-to-fine Semantic Segmentation from Image-level Labels
• Navigation Agents for the Visually Impaired: A Sidewalk Simulator and Experiments
• Deeply Cascaded U-Net for Multi-Task Image Processing
• Improving Robustness of Learning-based Autonomous Steering Using Adversarial Images
• Learning Debiased and Disentangled Representations for Semantic Segmentation
• Cross-Domain Document Object Detection: Benchmark Suite and Method
• A Network Structure to Explicitly Reduce Confusion Errors in Semantic Segmentation
• Dynamic Objects Segmentation for Visual Localization in Urban Environments
• Locally Adaptive Learning Loss for Semantic Image Segmentation
• Function Norms and Regularization in Deep Networks
• MultiNet: Multi-Modal Multi-Task Learning for Autonomous Driving
• m-RevNet: Deep Reversible Neural Networks with Momentum
• Instance and Panoptic Segmentation Using Conditional Convolutions
• DF-Net: Unsupervised Joint Learning of Depth and Flow using Cross-Task Consistency
• IDD: A Dataset for Exploring Problems of Autonomous Navigation in Unconstrained Environments
• Distilling portable Generative Adversarial Networks for Image Translation
• Adversarial Attacks on Co-Occurrence Features for GAN Detection
• A Projected Gradient Descent Method for CRF Inference allowing End-To-End Training of Arbitrary Pairwise Potentials
• Unsupervised Video Depth Estimation Based on Ego-motion and Disparity Consensus
• SS3D: Single Shot 3D Object Detector
• On Machine Learning and Structure for Mobile Robots
• GuidedMix-Net: Learning to Improve Pseudo Masks Using Labeled Images as Reference
• SegNBDT: Visual Decision Rules for Segmentation
• Unsupervised Domain Adaptation: from Simulation Engine to the RealWorld
• Improving Semantic Segmentation via Dilated Affinity
• Complex Urban LiDAR Data Set
• Normalization in Training U-Net for 2D Biomedical Semantic Segmentation
• Frame-To-Frame Consistent Semantic Segmentation
• Dual Attention GANs for Semantic Image Synthesis
• Vision-based Navigation of Autonomous Vehicle in Roadway Environments with Unexpected Hazards
• Instance-Level Task Parameters: A Robust Multi-task Weighting Framework
• Low-Latency Video Semantic Segmentation
• A Systematic Comparison of Deep Learning Architectures in an Autonomous Vehicle
• Boosting Image Outpainting with Semantic Layout Prediction
• Attentional Separation-and-Aggregation Network for Self-supervised Depth-Pose Learning in Dynamic Scenes
• Adversarial Pixel-Level Generation of Semantic Images
• FusionNet and AugmentedFlowNet: Selective Proxy Ground Truth for Training on Unlabeled Images
• A Methodological Review of Visual Road Recognition Procedures for Autonomous Driving Applications
• Contradictory Structure Learning for Semi-supervised Domain Adaptation
• AIBench Training: Balanced Industry-Standard AI Training Benchmarking
• Pixelwise Instance Segmentation with a Dynamically Instantiated Network
• Automatic generation of ground truth for the evaluation of obstacle detection and tracking techniques
• S2R-DepthNet: Learning a Generalizable Depth-specific Structural Representation
• CompressNet: Generative Compression at Extremely Low Bitrates
• Fast, Better Training Trick -- Random Gradient
• AIO-P: Expanding Neural Performance Predictors Beyond Image Classification
• Exemplar-Based Open-Set Panoptic Segmentation Network
• DSIC: Deep Stereo Image Compression
• A Fully Convolutional Tri-branch Network (FCTN) for Domain Adaptation
• Neural Architecture Search with Random Labels
• Generalizing to the Open World: Deep Visual Odometry with Online Adaptation
• Structural Consistency and Controllability for Diverse Colorization
• Learning by Inertia: Self-supervised Monocular Visual Odometry for Road Vehicles
• Simple multi-dataset detection
• Stereo 3D Object Trajectory Reconstruction
• Risk Bounds for Unsupervised Cross-Domain Mapping with IPMs
• DDD20 End-to-End Event Camera Driving Dataset: Fusing Frames and Events with Deep Learning for Improved Steering Prediction
• Unsupervised Metric Relocalization Using Transform Consistency Loss
• Importance of Self-Consistency in Active Learning for Semantic Segmentation
• Binary-decomposed DCNN for accelerating computation and compressing model without retraining
• Efficient Semantic Segmentation using Gradual Grouping
• Dynamic Video Segmentation Network
• Transferring and Regularizing Prediction for Semantic Segmentation
• EADNet: Efficient Asymmetric Dilated Network for Semantic Segmentation
• GANtruth - an unpaired image-to-image translation method for driving scenarios
• FixNorm: Dissecting Weight Decay for Training Deep Neural Networks
• Distance to Center of Mass Encoding for Instance Segmentation
• SILG: The Multi-environment Symbolic Interactive Language Grounding Benchmark
• ClimateGAN: Raising Climate Change Awareness by Generating Images of Floods
• One-Click Annotation with Guided Hierarchical Object Detection
• Interpretable Self-Attention Temporal Reasoning for Driving Behavior Understanding
• Weakly- and Semi-Supervised Panoptic Segmentation
• SMU: smooth activation function for deep networks using smoothing maximum technique
• Explainable Semantic Mapping for First Responders
• Inter-BMV: Interpolation with Block Motion Vectors for Fast Semantic Segmentation on Video
• Generating superpixels using deep image representations
• Alleviating Semantic-level Shift: A Semi-supervised Domain Adaptation Method for Semantic Segmentation
• SIGNet: Semantic Instance Aided Unsupervised 3D Geometry Perception
• Generating Easy-to-Understand Referring Expressions for Target Identifications
• Improved Semantic Stixels via Multimodal Sensor Fusion
• A Domain Agnostic Normalization Layer for Unsupervised Adversarial Domain Adaptation
• Scene Parsing via Dense Recurrent Neural Networks with Attentional Selection
• Composite Binary Decomposition Networks
• Recurrent Flow-Guided Semantic Forecasting
• Scale Calibrated Training: Improving Generalization of Deep Networks via Scale-Specific Normalization
• StuffNet: Using 'Stuff' to Improve Object Detection
• A perception centred self-driving system without HD Maps
• Predicting Ground-Level Scene Layout from Aerial Imagery
• DFNet: Semantic Segmentation on Panoramic Images with Dynamic Loss Weights and Residual Fusion Block
• A Large-Scale Benchmark for Food Image Segmentation
• Excavate Condition-invariant Space by Intrinsic Encoder
• Dense Prediction with Attentive Feature Aggregation
• Verisimilar Percept Sequences Tests for Autonomous Driving Intelligent Agent Assessment
• Reducing the feature divergence of RGB and near-infrared images using Switchable Normalization
• Semantic Segmentation and Object Detection Towards Instance Segmentation: Breast Tumor Identification
• Exploiting Inter-pixel Correlations in Unsupervised Domain Adaptation for Semantic Segmentation
• The NEOLIX Open Dataset for Autonomous Driving
• Bidirectional Multi-scale Attention Networks for Semantic Segmentation of Oblique UAV Imagery
• End-to-End Pedestrian Collision Warning System based on a Convolutional Neural Network with Semantic Segmentation
• Zero-shot Adversarial Quantization
• Memory-guided Unsupervised Image-to-image Translation
• Taming Adversarial Domain Transfer with Structural Constraints for Image Enhancement
• Future Video Synthesis with Object Motion Prediction
• Long-Term On-Board Prediction of People in Traffic Scenes under Uncertainty
• What Can Help Pedestrian Detection?
• Forecasting Hands and Objects in Future Frames
• Cross-View Image Synthesis with Deformable Convolution and Attention Mechanism
• Multimodal Aggregation Approach for Memory Vision-Voice Indoor Navigation with Meta-Learning
• A New Ensemble Adversarial Attack Powered by Long-term Gradient Memories
• Multichannel Semantic Segmentation with Unsupervised Domain Adaptation
• Unsupervised Video Object Segmentation with Distractor-Aware Online Adaptation
• Unsupervised Learning-based Depth Estimation aided Visual SLAM Approach
• Region Deformer Networks for Unsupervised Depth Estimation from Unconstrained Monocular Videos
• One-Shot Image-to-Image Translation via Part-Global Learning with a Multi-adversarial Framework
• Pedestrian Prediction by Planning using Deep Neural Networks
• Robust Multi-Image HDR Reconstruction for the Modulo Camera
• weedNet: Dense Semantic Weed Classification Using Multispectral Images and MAV for Smart Farming
• Robust Deep-Learning-Based Road-Prediction for Augmented Reality Navigation Systems
• Functionally Modular and Interpretable Temporal Filtering for Robust Segmentation
• From Virtual to Real World Visual Perception using Domain Adaptation -- The DPM as Example
• Safety Metrics for Semantic Segmentation in Autonomous Driving
• Video Class Agnostic Segmentation with Contrastive Learning for Autonomous Driving
• Learning Dilation Factors for Semantic Segmentation of Street Scenes
• Split-Merge Pooling
• Fully Convolutional Neural Networks for Dynamic Object Detection in Grid Maps (Masters Thesis)
• Reproducibility of "FDA: Fourier Domain Adaptation forSemantic Segmentation
• MASON: A Model AgnoStic ObjectNess Framework
• Advanced Mapping Robot and High-Resolution Dataset
• Collaborative Annotation of Semantic Objects in Images with Multi-granularity Supervisions
• Accurate and efficient video de-fencing using convolutional neural networks and temporal information
• Vehicle Image Generation Going Well with The Surroundings
• Efficient Semantic Segmentation for Visual Bird's-eye View Interpretation
• Noisy-LSTM: Improving Temporal Awareness for Video Semantic Segmentation
• Superpixel Sampling Networks
• Learning Propagation for Arbitrarily-structured Data
• Scene Completeness-Aware Lidar Depth Completion for Driving Scenario
• Affinity Derivation and Graph Merge for Instance Segmentation
• CrDoCo: Pixel-level Domain Transfer with Cross-Domain Consistency
• Object Detection on Single Monocular Images through Canonical Correlation Analysis
• Per-Pixel Feedback for improving Semantic Segmentation
• Instance Map based Image Synthesis with a Denoising Generative Adversarial Network
• Unsupervised Image-to-Image Translation with Stacked Cycle-Consistent Adversarial Networks
• Accelerated Inference in Markov Random Fields via Smooth Riemannian Optimization
• Joint Optical Flow and Temporally Consistent Semantic Segmentation
• Robustness via Uncertainty-aware Cycle Consistency
• RethNet: Object-by-Object Learning for Detecting Facial Skin Problems
• Semantic Segmentation with Scarce Data
• Collaging Class-specific GANs for Semantic Image Synthesis
• Instance-Conditioned GAN
• Learning Deep Multimodal Feature Representation with Asymmetric Multi-layer Fusion
• Unsupervised Learning of Depth and Deep Representation for Visual Odometry from Monocular Videos in a Metric Space
• Merging Tasks for Video Panoptic Segmentation
• Safe Augmentation: Learning Task-Specific Transformations from Data
• Hierarchical Pyramid Representations for Semantic Segmentation
• SMET: Scenario-based Metamorphic Testing for Autonomous Driving Models
• Semantically Guided Depth Upsampling
• Domain Adaptive Monocular Depth Estimation With Semantic Information
• A Brief Survey and an Application of Semantic Image Segmentation for Autonomous Driving
• A De-raining semantic segmentation network for real-time foreground segmentation
• Heterogeneous Grid Convolution for Adaptive, Efficient, and Controllable Computation
• On Offline Evaluation of Vision-based Driving Models
• Euro-PVI: Pedestrian Vehicle Interactions in Dense Urban Centers
• ContourRender: Detecting Arbitrary Contour Shape For Instance Segmentation In One Pass
• Non-local RoIs for Instance Segmentation
• Attention Guided Cosine Margin For Overcoming Class-Imbalance in Few-Shot Road Object Detection