3D generic object categorization, localization and pose estimation

By Silvio Savarese, Li Fei-Fei. 2007

455 citations recorded by the source corpus.

Subfields: Aerospace Engineering, 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 Aerospace Engineering

In Computer Vision and Pattern Recognition

Citations in the corpus

End-to-End Training of Deep Visuomotor Policies (2015, methodology): “…window to localize the object, reducing the task to classiﬁcation [8], perform regression to a heatmap of manually labeled keypoints [40], requiring precise knowledge of the object position in the image and camera calibration, or use 3D models to localize previously scanned objects [30, 36].”
Pix3D: Dataset and Methods for Single-Image 3D Shape Modeling (2018, background): “Many of the aforementioned object datasets include annotations of object poses [34, 38, 47, 63, 64].”
3D Shape Induction from 2D Views of Multiple Objects (2016, methodology): “Recognition-based methods have been used to estimate geometry of out-door scenes [14, 26], indoor environments [9, 27], and objects [1, 25].”
Render for CNN: Viewpoint Estimation in Images Using CNNs Trained with Rendered 3D Model Views (2015, methodology): “Compared with traditional coarse-grained classification-based formulations that typically have 8 to 24 discrete classes [29, 39], our formulation is capable of producing much more finegrained viewpoint estimation.”
Viewpoints and keypoints (2014, background): “Previous works [32, 15, 16] have also jointly tackled the problem of keypoint detection and pose estimation.”
Learning 3D Object Categories by Looking Around Them (2017, background): “The vast majority of methods for learning the viewpoint of object categories use manual supervision [32, 25, 11, 26, 43, 23, 39] or synthetic [36] data.”
Multiple Viewpoint Recognition and Localization (2010, unknown): “The training and validation data we used for the shoe classiﬁer came primarily from Savarese et al. [16], with a few additional images acquired from internet imagery.”
Towards Viewpoint Invariant 3D Human Pose Estimation (2016, background): “Given the success of SIFT [44], there have been several attempts at embedding rotational and translational invariance [55,62,2].”
Sparse distance learning for object recognition combining RGB and depth information (2011, methodology): “this dataset using the same setup as [22]: we randomly select 7 instances per category for training and use the rest for testing.”
Enriching object detection with 2D-3D registration and continuous viewpoint estimation (2015, methodology): “Figure 7: Detection results on 3D Object Classes [21] .”
A multi-view probabilistic model for 3D object classes (2009, methodology): “Unlike [ 30], but similar to [14, 25, 26] our model parts are linked across views under affine transformations.”
Convolutional Models for Joint Object Categorization and Pose Estimation (2015, background): “Recently, detection and pose have been solved simultaneously (e.g.(Tulsiani & Malik, 2014; Pepik et al., 2015b; Xiang et al., 2014; Savarese & Fei-Fei, 2007; Savarese & Li, 2008; Mei et al., 2011; Payet & Todorovic, 2011; Schels et al., 2012; Pepik et al., 2012; Lai et al., 2011b)).”
World-scale mining of objects and events from community photo collections (2008, background): “First, more powerful (local) visual features [3, 14, 15, 16, 30] have led to signiﬁcant progress in recognition capabilities, both for speciﬁc objects [27] and for object classes [7, 12, 24].”
Depth-Encoded Hough Voting for Joint Object Detection and Shape Recovery (2010, methodology): “Examples of the complete 3D object inference process using the testing images from Pascal VOC07 [6], ETHZ Shape [9], and 3D object dataset [21].”
Single image 3D object detection and pose estimation for grasping (2014, background): “Savarese and Fei-Fei [20] presented an approach for object categorization that combines appearance-based descriptors including the canonical view for each part, and transformations between parts.”
Learning descriptors for object recognition and 3D pose estimation (2015, background): “[30] pioneered this approach, and learned canonical parts connected by a graph for object recognition and pose estimation.”