Back to the Future: Learning Shape Models from 3D CAD Data

By Michael Stark, Michael Goesele, Bernt Schiele. 2010

178 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, 3D Vision & Pose

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In 3D Vision & Pose

Citations in the corpus

Domain-Adversarial Training of Neural Networks (2015, background): “…example is training an image classiﬁer on synthetic or semi-synthetic images, which may come in abundance and be fully labeled, but which inevitably have a distribution that is diﬀerent from real images (Liebelt and Schmid, 2010; Stark et al., 2010; V´azquez et al., 2014; Sun and Saenko, 2014).”
Playing for Data: Ground Truth from Computer Games (2016, neutral): “3D object models are also increasingly used for training representations for object detection and object pose estimation [4, 26, 30, 34, 35, 45].”
Render for CNN: Viewpoint Estimation in Images Using CNNs Trained with Rendered 3D Model Views (2015, methodology): “Recently, [32, 19, 13, 27] used 3D models to render images for training object detectors and viewpoint classifiers.”
Domain Adaptation for Visual Applications: A Comprehensive Survey (2017, methodology): “Early methods used 3D CAD models to improve solutions for pose and viewpoint estimation [2, 205, 162, 214], object and object part detection [213, 166, 217, 182, 165, 112, 145, 16], segmentation and scene understanding [187, 33, 163].”
VirtualWorlds as Proxy for Multi-object Tracking Analysis (2016, background): “First attempts to use synthetic data for training are mainly limited to using rough synthetic models or synthesized real examples ( e.g., of pedestrians [14, 15]).”
Learning Non-Lambertian Object Intrinsics Across ShapeNet Categories (2016, methodology): “Images rendered from 3D models are widely used in deep learning, e.g. [31, 19, 13, 23] for training object detectors and viewpoint classiﬁers.”
From Virtual to Reality: Fast Adaptation of Virtual Object Detectors to Real Domains (2014, background): “Most previous work utilizing 3D models focuses on predicting the 3D pose or viewpoint of an object in 2D test images [14, 18, 22, 23, 25], or categorizing an object from an arbitrary novel viewpoint [21].”
3D Object Detection and Viewpoint Estimation with a Deformable 3D Cuboid Model (2012, methodology): “Since these methods usually require a significant amount of training data, renderings of synthetic CAD models have been used to supplement under-represented views or provide supervision for training object parts or object geometry [22, 13, 8].”
RotationNet: Joint Object Categorization and Pose Estimation Using Multiviews from Unsupervised Viewpoints (2016, background): “ signiﬁcance of object pose estimation from a single-view image is also discussed in recent publications that synthesize 3D CAD models of objects into real images for learning real object recognition [16,17,13,18,19,20]. Papon and Schoeler [16] learned a CNN that takes the normal, intensity, and dep”
Model Based Training, Detection and Pose Estimation of Texture-Less 3D Objects in Heavily Cluttered Scenes (2012, methodology): “To overcome these limitations, researchers tried to learn the object appearance from 3D models [7, 8, 10].”
Configurable 3D Scene Synthesis and 2D Image Rendering with Per-pixel Ground Truth Using Stochastic Grammars (2017, background): “surface normal prediction, semantic segmentation, reconstruction, etc. 1.1 Related Work Synthetic image datasets have recently been a source of trainingdataforobjectdetectionandcorrespondencematching [26, 32, 37, 83, 91, 95, 113, 117, 120, 150], single-view reconstruction [58], view-point estimation”
Multimodal templates for real-time detection of texture-less objects in heavily cluttered scenes (2011, background): “As such, they work even when the object is not textured enough to use feature point techniques and can directly provide a coarse estimation of the object pose.”
Teaching 3D geometry to deformable part models (2012, methodology): “For i), we follow [28] to generate a non-photorealistic, gradientbased rendering of the 3D CAD model, and extract a HOG filter for the part p j directly from that rendering.”
Play and Learn: Using Video Games to Train Computer Vision Models (2016, methodology): “There are also methods that rely on indirect representations such as the HOG templates [16] or geometrical descriptors [23, 42], that can be efficiently synthesized and used.”
Scenic: a language for scenario specification and scene generation (2018, background): “rrect differencesbetweensyntheticandreal-worldinputdistributions.Domain adaptation has enabled synthetic data to successfully train models for several other applications including 3D object detection [29, 46], pedestrian detection [50], and semantic image segmentation [42]. Such work provides import”