From contours to 3D object detection and pose estimation

Payet, Nadia; Todorovic, Sinisa

doi:10.1109/iccv.2011.6126342

Cited by 115 publications

(98 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To overcome these limitations, researchers tried to learn the object appearance from 3D models [7,8,10]. The approach of Stark et al [7] relies only on 3D CAD models of cars and Liebelt and Schmid [8] combine geometric shape and pose priors with natural images.…”

Section: Related Workmentioning

confidence: 99%

“…Both of these approaches work well and also generalize to object classes, but they are not real-time capable, require expensive training and cannot handle clutter and occlusions well. In [10] authors use a number of viewpoint-specific shape representations to model the object category. They rely on contours and introduce a novel feature called BOB (bag of boundaries), which at a given point in the image is a histogram of boundaries from image contours in training images.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Model Based Training, Detection and Pose Estimation of Texture-Less 3D Objects in Heavily Cluttered Scenes

Hinterstoißer

Lepetit

Ilić

et al. 2013

Lecture Notes in Computer Science

916

1,298

View full text Add to dashboard Cite

Abstract. We propose a framework for automatic modeling, detection, and tracking of 3D objects with a Kinect. The detection part is mainly based on the recent template-based LINEMOD approach [1] for object detection. We show how to build the templates automatically from 3D models, and how to estimate the 6 degrees-of-freedom pose accurately and in real-time. The pose estimation and the color information allow us to check the detection hypotheses and improves the correct detection rate by 13% with respect to the original LINEMOD. These many improvements make our framework suitable for object manipulation in Robotics applications. Moreover we propose a new dataset made of 15 registered, 1100+ frame video sequences of 15 various objects for the evaluation of future competing methods. Fig. 1. 15 different texture-less 3D objects are simultaneously detected with our approach under different poses on heavy cluttered background with partial occlusion. Each detected object is augmented with its 3D model. We also show the corresponding coordinate systems.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Model Based Training, Detection and Pose Estimation of Texture-Less 3D Objects in Heavily Cluttered Scenes

Hinterstoißer

Lepetit

Ilić

et al. 2013

Lecture Notes in Computer Science

916

1,298

View full text Add to dashboard Cite

show abstract

“…Common approaches to deal with the problem of detecting shapes address these issues by relying on shape information estimated upon gray-scale images, i.e., by extracting contours and local orientations based on the local gradients of intensity images rather than using binary images [12]. For instance, using a networks of local segment as descriptors, and performing detection of shapes belonging to classes that are relatively easy to differentiate in visual terms [13].…”

Section: Related Workmentioning

confidence: 99%

Evaluating Shape Descriptors for Detection of Maya Hieroglyphs

Roman-Rangel

Odobez

Gática-Pérez

2013

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. In this work we address the problem of detecting instances of complex shapes in binary images. We investigated the effects of combining DoG and Harris-Laplace interest points with SIFT and HOOSC descriptors. Also, we propose the use of a retrieval-based detection framework suitable to deal with images that are sparsely annotated, and where the objects of interest are very small in proportion to the total size of the image. Our initial results suggest that corner structures are suitable points to compute local descriptors for binary images, although there is the need for better methods to estimate their appropriate characteristic scale when used on binary images.

show abstract

“…Testing is performed on scenes ("Table-Top-Local") containing one or several instances of the objects in a cluttered office environment; note that those experimental conditions are more challenging than existing evaluations (e.g. in [56]) since those two parts of the dataset feature different imaging and lighting conditions. We perform detection in the test images of each object category separately, and we measure the detection rates with the standard criterion of 50% bounding box overlap.…”

Section: Tabletop Dataset: Multiview Model Detection In Cluttermentioning

confidence: 99%

Multiview feature distributions for object detection and continuous pose estimation

Teney

Piater

2014

Computer Vision and Image Understanding

View full text Add to dashboard Cite

This paper presents a multiview model of object categories, generally applicable to virtually any type of image features, and methods to efficiently perform, in a unified manner, detection, localization and continuous pose estimation in novel scenes. We represent appearance as distributions of low-level, fine-grained image features. Multiview models encode the appearance of objects at discrete viewpoints, and, in addition, how these viewpoints deform into one another as the viewpoint continuously varies (as detected from optical flow between training examples). Using a measure of similarity between an arbitrary test image and such a model at chosen viewpoints, we perform all tasks mentioned above with a common method. We leverage the simplicity of low-level image features, such as points extracted along edges, or coarse-scale gradients extracted densely over the images, by building probabilistic templates, i.e. distributions of features, learned from one or several training examples. We efficiently handle these distributions with probabilistic techniques such as kernel density estimation, Monte Carlo integration and importance sampling. We provide an extensive evaluation on a wide variety of benchmark datasets. We demonstrate performance on the "ETHZ Shape" dataset, with single (hand-drawn) and multiple training examples, well above baseline methods, on par with a number of more task-specific methods. We obtain remarkable performance on the recognition of more complex objects, notably the cars of the "3D Object" dataset of Savarese et al. with detection rates of 92.5% and an accuracy in pose estimation of 91%. We perform better than the state-of-the-art on continuous pose estimation with the "rotating cars" dataset of Ozuysal et al. . We also demonstrate particular capabilities with a novel dataset featuring non-textured objects of undistinctive shapes, the pose of which can only be determined from shading, captured here by coarse scale intensity gradients.

show abstract

From contours to 3D object detection and pose estimation

Cited by 115 publications

References 14 publications

Model Based Training, Detection and Pose Estimation of Texture-Less 3D Objects in Heavily Cluttered Scenes

Model Based Training, Detection and Pose Estimation of Texture-Less 3D Objects in Heavily Cluttered Scenes

Evaluating Shape Descriptors for Detection of Maya Hieroglyphs

Multiview feature distributions for object detection and continuous pose estimation

Contact Info

Product

Resources

About