Parsing IKEA Objects: Fine Pose Estimation

By Joseph J. Lim, Hamed Pirsiavash, Antonio Torralba. 2013

307 citations recorded by the source corpus.

We address the problem of localizing and estimating the fine-pose of objects in the image with exact 3D models. Our main focus is to unify contributions from the 1970s with recent advances in object detection: use local keypoint detectors to find candidate poses and score global alignment of each candidate pose to the image. Moreover, we also provide a new dataset containing fine-aligned objects with their exactly matched 3D models, and a set of models for widely used objects. We also evaluate our algorithm both on object detection and fine pose estimation, and show that our method outperforms state-of-the art algorithms.

Subfields: Aerospace Engineering, 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 Aerospace Engineering

In Computer Vision and Pattern Recognition

In 3D Vision & Pose

Citations in the corpus

Fine-Grained Egocentric Hand-Object Segmentation: Dataset, Model, and Applications (2022, background): “One highly related direction is to model and estimate 3D hand joints [5,44,45,53,60,63,66,72,78,79,80,85,3] and mesh [26,27,51,55,75,83,46], object pose [14,28,29,35,43,57,64,71,76,81], or both [6,17,18,19,69].”
Pix3D: Dataset and Methods for Single-Image 3D Shape Modeling (2018, methodology): “Another closely related benchmark is IKEA [38], which provides accurate alignment between images of IKEA objects and 3D CAD models.”
Learning a Probabilistic Latent Space of Object Shapes via 3D Generative-Adversarial Modeling (2016, methodology): “The discriminator basically mirrors the generator, except that it uses Leaky ReLU [Maas et al., 2013] instead of ReLU layers.”
Learning a Predictable and Generative Vector Representation for Objects (2016, methodology): “Many approaches achieve this via alignment [1, 14, 22]; the most similar work is Li et al.”
Lifting Object Detection Datasets into 3D (2015, positive): “To account for differences between the 3D exemplars and the object depicted, Su and Guibas [26] choose a few exemplars that appear to be similar to the depicted shape and combine them to produce a single depth map.”
Category-specific object reconstruction from a single image (2014, methodology): “Polyhedral models [18, 40], which trace back to the early work of Roberts [29], and CAD models [25, 31] provide crude approximations of shape and given a set of point correspondences can be quite effective for determining instance viewpoints.”
Single Image 3D Interpreter Network (2016, methodology): “We use the IKEA dataset [14] for evaluation, as it provides ground truth 3D mesh models and the associated viewpoints for testing images.”
Render for CNN: Viewpoint Estimation in Images Using CNNs Trained with Rendered 3D Model Views (2015, methodology): “For example, [32] and [20] proposed similar methods that learn a 3D deformable part model and demonstrate superior performance for cars and chairs, respectively; [21] and [1] formulated an alignment problem and built key point correspondences between 2D images and rendered 3D views.”
Monocular Object Instance Segmentation and Depth Ordering with CNNs (2015, methodology): “Online object catalogs can be used to assign physical size to the CAD models based on their semantic class, giving rise to a 3D interpretation of the scene.”
Marr Revisited: 2D-3D Alignment via Surface Normal Prediction (2016, background): “Recent work include instance detection of a small set of IKEA objects via contour-based alignment [26], depth prediction by aligning to renders of 3D shapes via hand-crafted features [44], object class detection via exemplar matching with mid-level elements [1, 7], and alignment via composition from”
Past, Present, and Future of Simultaneous Localization and Mapping: Toward the Robust-Perception Age (2016, neutral): “sentations, which define a solid as a combination of atoms in a dictionary, have been considered in robotics and computer vision, with dictionary learned from data [266] or based on existing repositories of object models [149], [157].”
Instance-Level Segmentation for Autonomous Driving with Deep Densely Connected MRFs (2015, background): “Another line of work matches CAD models to images [2, 20, 10].”
Learning 3D Object Categories by Looking Around Them (2017, background): “Often the idea is to search for the 3D model in a CAD library that best fits the image [19, 1, 13, 2].”
Adaptive region pooling for object detection (2015, background): “Non-rigid object models [19] and 3D CAD models [3, 28] can also be used to generalize the application of transferring rigid and non-rigid object information.”
Completing 3D object shape from one depth image (2015, background): “Automatic and semi-automatic approaches exist for chairs [2] and furniture [23, 39] that have highly informative contours and an enormous supply of available 3D models.”
T-LESS: An RGB-D Dataset for 6D Pose Estimation of Texture-Less Objects (2017, methodology): “[30] provides RGB images with objects being aligned with their exactly matched 3D models.”