From Virtual to Reality: Fast Adaptation of Virtual Object Detectors to Real Domains
Explore this paper's citation graph
Summary
This work investigates the use of such freely available 3D models for multicategory 2D object detection and proposes a simple and fast adaptation approach based on decorrelated features, which performs comparably to existing methods trained on large-scale real image domains.
- Type
- article
- Published
- 2014-01-01
- Cited by
- 170
- References
- 31
- Access
- Open access
- OpenAlex
- https://openalex.org/W2083544878
- Semantic Scholar
- https://api.semanticscholar.org/CorpusID:6018652
Keywords
Computer science, Object detection, Benchmark (surveying), Virtual reality, Artificial intelligence
References
- Efficient Learning of Domain-invariant Image Representations
- Teaching 3D geometry to deformable part models
- Cross-domain video concept detection using adaptive svms
- Tabula rasa: Model transfer for object category detection
- Ensemble of exemplar-SVMs for object detection and beyond
- Beyond PASCAL: A benchmark for 3D object detection in the wild
- Incremental Domain Adaptation of Deformable Part-based Models
- Adapting a Pedestrian Detector by Boosting LDA Exemplar Classifiers
- Unbiased look at dataset bias
- The Pascal Visual Object Classes (VOC) Challenge
- Virtual and Real World Adaptation for Pedestrian Detection
- Domain Adaptation of Deformable Part-Based Models
- Enhancing Exemplar SVMs using Part Level Transfer Regularization
- Interactive adaptation of real-time object detectors
- What you saw is not what you get: Domain adaptation using asymmetric kernel transforms
- Description and Recognition of Curved Objects
- ImageNet: A large-scale hierarchical image database
- Viewpoint-independent object class detection using 3D Feature Maps
- Domain adaptation for object recognition: An unsupervised approach
- A multi-view probabilistic model for 3D object classes
Cited by
- A Unified Perspective on Multi-Domain and Multi-Task Learning
- Exploring Invariances in Deep Convolutional Neural Networks Using Synthetic Images
- Learning scene-specific pedestrian detectors without real data
- Salient Object Subitizing
- See the Difference: Direct Pre-Image Reconstruction and Pose Estimation by Differentiating HOG
- Return of Frustratingly Easy Domain Adaptation
- Towards Adapting Deep Visuomotor Representations from Simulated to Real Environments
- Deep Exemplar 2D-3D Detection by Adapting from Real to Rendered Views
- Learning from Synthetic Data Using a Stacked Multichannel Autoencoder
- Scatter Component Analysis: A Unified Framework for Domain Adaptation and Domain Generalization
- Fine-grain uncommon object detection from satellite images
- Subspace Distribution Alignment for Unsupervised Domain Adaptation
- Human Pose as Context for Object Detection
- VirtualWorlds as Proxy for Multi-object Tracking Analysis
- The SYNTHIA Dataset: A Large Collection of Synthetic Images for Semantic Segmentation of Urban Scenes
- Deep CORAL: Correlation Alignment for Deep Domain Adaptation
- Play and Learn: Using Video Games to Train Computer Vision Models
- Domain Adaptation and Domain Generalization with Representation Learning
- High-precision bicycle detection on single side-view image based on the geometric relationship
- Using 3D Graphics to Train Object Detection Systems
Related papers
- Bounding-box Centralization for Improving SiamFC++
- Syncretic-NMS: A Merging Non-Maximum Suppression Algorithm for Instance Segmentation
- Medical image segmentation with imperfect 3D bounding boxes
- A Method to Generate the Minimum Bounding Boxes for Shape-Arbitrary Objects
- Fully Functional Image Manipulation Using Scene Graphs in A Bounding-Box Free Way
- Bottom-up Pose Estimation of Multiple Person with Bounding Box Constraint
- Shape-IoU: More Accurate Metric considering Bounding Box Shape and Scale