The Princeton Shape Benchmark

By Philip Shilane, P. Min, M. Kazhdan, Thomas A. Funkhouser. 2004

1,708 citations recorded by the source corpus.

Subfields: Computational Mechanics, Computer Vision and Pattern Recognition

How researchers see this paper

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

In Computational Mechanics

In Computer Vision and Pattern Recognition

Citations in the corpus

ShapeNet: An Information-Rich 3D Model Repository (2015, methodology): “The Princeton Shape Benchmark is probably the most well-known and frequently used 3D shape collection to date (with over 1000 citations) [27].”
Pix3D: Dataset and Methods for Single-Image 3D Shape Modeling (2018, methodology): “For decades, researchers have been building datasets of 3D objects, either as a repository of 3D CAD models [4, 5, 49] or as images of 3D shapes with pose annotations [34, 47].”
Semantically-enriched 3D models for common-sense knowledge (2015, background): “Absolute size is critical in the real world since it influences the usability of objects and even their identity (e.g., a model airplane vs a real airplane).”
Supervised learning of hidden and non-hidden 0-order affordances and detection in real scenes (2012, methodology): “Princeton Shape Benchmark [23] models displayed in the pose where, according to our approach, the sittable affordance was detected as fulfilled (detected using a SVM classifier with a PFH descriptor).”
Open X-Embodiment: Robotic Learning Datasets and RT-X Models : Open X-Embodiment Collaboration0 (2023, positive): “…we directly train a policy on X-embodiment data, without any mechanisms to reduce the embodiment gap, and observe positive transfer by leveraging that data. spanning grasping [60–71], pushing interactions [23, 72–74], sets of objects and models [75–85], and teleoperated demonstrations [8, 86–95].”
Context-based search for 3D models (2010, methodology): “A 3D model database that contains only isolated objects, such as the Princeton Shape Benchmark, provides no information about the relationships between objects in real scenes [Shilane et al. 2004].”
3D Object Recognition in Cluttered Scenes with Local Surface Features: A Survey (2014, neutral): “2 PSB [58] Mesh 1814 Not Available”
Recent Trends, Applications, and Perspectives in 3D Shape Similarity Assessment (2016, methodology): “…similarity are the Princeton Shape Benchmark , conceived at ﬁrst for shape retrieval and now extended to shape correspondence and segmentation [SMKF04, CGF09]; the McGill 3D Shape Benchmark , in part building on the Princeton Shape Benchmark but speciﬁcally designed for non-rigid shape…”
Deepshape: Deep learned shape descriptor for 3D shape matching and retrieval (2015, background): “Shape descriptors for shape matching and retrieval have been extensively studied in the geometry community [30, 14, 12, 31, 25].”
Shrec'16 Track: Retrieval of Human Subjects from Depth Sensor Data (2016, methodology): “The retrieval performance of baseline and participants’ meth-ods were evaluated according to the classical measures used in [SMKF04], e.g. Nearest Neighbor (NN), First Tier (FT), Second Tier (ST), e-measure (E) and Discounted Cumualted Gain (DCG).”
Benchmarking in Manipulation Research: The YCB Object and Model Set and Benchmarking Protocols (2015, methodology): “24 The Princeton Shape Benchmark [18] 2004 Mesh models Shape-based retrieval 1,814 No http://shape.”
Performance Evaluation of 3D Keypoint Detectors (2012, neutral): “An alternative approach, generally adopted in the absence of clutter and occlusions, relies instead on describing the whole surface by means of a single descriptor (Shang and Greenspan 2010; Akgül et al. 1992; Shilane et al. 2004).”
Hough Transform and 3D SURF for Robust Three Dimensional Classification (2010, methodology): “8K shapes (half training, half testing), 7 classes taken from the Princeton Benchmark [24].”
Curve-Skeleton Properties, Applications, and Algorithms (2007, methodology): “In [113], some of the best performing shape descriptors for shape matching cannot reconstruct the object at all.”
Salient local visual features for shape-based 3D model retrieval (2008, methodology): “To evaluate retrieval performance for rigid models, we used the Princeton Shape Benchmark (PSB) [30].”
3D ShapeNets: A deep representation for volumetric shapes (2014, methodology): “For comparison, we choose Light Field descriptor [8] (LFD, 4,700 dimensions) and Spherical Harmonic descriptor [18] (SPH, 544 dimensions), which performed best among all descriptors [28].”
A large-scale Shape Benchmark for 3D object retrieval: Toyohashi shape benchmark (2012, methodology): “If we can retrieve a 3D object that perfectly meets our need from massive 3D databases and integrate the retrieved object into a composite and complex 3D object, it is expected that we can greatly cut the expense of creating a new 3D object from scratch [3].”