research:appearance-models

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision
Previous revision
Next revision Both sides next revision
research:appearance-models [2013/07/13 09:55]
c7031009
research:appearance-models [2013/07/16 19:46]
c7031007
Line 1: Line 1:
 ====== Probabilistic Models of Appearance for Object Recognition and Pose Estimation ====== ====== Probabilistic Models of Appearance for Object Recognition and Pose Estimation ======
  
-{{ :​research:​teney-2013-crv4.jpg?​nolink&​300}} We developped ​probabilistic models to encode the appearance of objects, and inference methods to perform **detection (localization) and pose estimation** of those object **in 2D images** of cluttered scenes. Some of our early work used 3D, CAD-style models ([[@/​publications#​Teney-2011-DAGM|Teney et al. 2011]]), but we then solely focused on **appearance-based models** ([[@/​publications#​Teney-2012-DICTA|Teney et al. 2012]]). Those are trained using 2D example images alone, the goal being here to leverage, to a maximum, the information conveyed by 2D images, without resorting to stereo or other 3D sensing techniques. Our models are identically applicable to either specific object instances, or to object categories/​classes ([[@/​publications#​Teney-2013-CRV|Teney et al. 2013]]). The appearance is modeled as a **distributions ​of low-level, fine-grained image features**. The strength of the approach is its straightforward formulation,​ applicable to virtually any type of image feature. We have applied the method to different types of such low-level features: points along image edges, and intensity gradients extracted densely over the image.+{{ :​research:​teney-2013-crv4.jpg?​nolink&​300}} We developed ​probabilistic models to encode the appearance of objects, and inference methods to perform **detection (localization) and pose estimation** of those object **in 2D images** of cluttered scenes. Some of our early work used 3D, CAD-style models ([[@/​publications#​Teney-2011-DAGM|Teney et al. 2011]]), but we then solely focused on **appearance-based models** ([[@/​publications#​Teney-2012-DICTA|Teney et al. 2012]]). Those are trained using 2D example images alone, the goal being here to leverage, to a maximum, the information conveyed by 2D images, without resorting to stereo or other 3D sensing techniques. Our models are identically applicable to either specific object instances, or to object categories/​classes ([[@/​publications#​Teney-2013-CRV|Teney et al. 2013]]). The appearance is modeled as a **distribution ​of low-level, fine-grained image features**. The strength of the approach is its straightforward formulation,​ applicable to virtually any type of image feature. We have applied the method to different types of such low-level features: points along image edges, and intensity gradients extracted densely over the image.
  
 <​html><​div style="​clear:​both"></​div><​br></​html>​ <​html><​div style="​clear:​both"></​div><​br></​html>​
Line 15: Line 15:
 <​html><​div style="​clear:​both"></​div><​br></​html>​ <​html><​div style="​clear:​both"></​div><​br></​html>​
  
-{{ :​research:​teney-2013-ip.jpg?​nolink&​300}} We found an interesting application of those models and methods in the **recognition,​ pose estimation and segmentation of a robotic arm in 2D images** ([[@/​publications#​Teney-2013-IP|Teney et al. 2013b]]). This task is very challenging due to the smooth and untextured appearance of the robot arm considered (a Kuka LWR). Moreover, the arm is made of articulated links which are absolutely identical in shape and appearance. Candidate detections of those links in the image are provided by the recognition method, and the known physical (kinematic) constraints between the articulated links are enforced by probabilistic inference. ​In the spirit of traditional articulated models, those constraints are modeled in a Markov random field, and an algorithm ​such as belief propagation can then identify a globally consistent result for the pose of all links. ​+{{ :​research:​teney-2013-ip.jpg?​nolink&​300}} We found an interesting application of these models and methods in the **recognition,​ pose estimation and segmentation of a robotic arm in 2D images** ([[@/​publications#​Teney-2013-IP|Teney et al. 2013b]]). This task is very challenging due to the smooth and untextured appearance of the robot arm considered (a Kuka LWR). Moreover, the arm is made of articulated links which are absolutely identical in shape and appearance. Candidate detections of those links in the image are provided by the recognition method, and the known physical (kinematic) constraints between the articulated links are enforced by probabilistic inference. ​Similarly to the traditional articulated models, those constraints are modeled in a Markov random field, and an algorithm ​based on belief propagation can then identify a globally consistent result for the configuration ​of all links. ​
research/appearance-models.txt · Last modified: 2018/09/03 19:35 (external edit)