Depth-Encoded Hough Voting for Joint Object Detection and Shape Recovery

Abstract

Abstract. Detecting objects, estimating their pose and recovering 3D shape information are critical problems in many vision and robotics applications. This paper addresses the above needs by proposing a new method called DEHV -Depth-Encoded Hough Voting detection scheme. Inspired by the Hough voting scheme introduced in [13], DEHV incorporates depth information into the process of learning distributions of image features (patches) representing an object category. DEHV takes advantage of the interplay between the scale of each object patch in the image and its distance (depth) from the corresponding physical patch attached to the 3D object. DEHV jointly detects objects, infers their categories, estimates their pose, and infers/decodes objects depth maps from either a single image (when no depth maps are available in testing) or a single image augmented with depth map (when this is available in testing). Extensive quantitative and qualitative experimental analysis on existing datasets [6,9,22] and a newly proposed 3D table-top object category dataset shows that our DEHV scheme obtains competitive detection and pose estimation results as well as convincing 3D shape reconstruction from just one single uncalibrated image. Finally, we demonstrate that our technique can be successfully employed as a key building block in two application scenarios (highly accurate 6 degrees of freedom (6 DOF) pose estimation and 3D object modeling).

Ensembles of strong learners for multi-cue classification

“…They showed that color-based cues are more important for instance recognition, while geometric ones are better suited for categorization, and that their combination improves on both. This finding is also supported by Sun et al (2010).…”

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

“…In recent years, a number of methods that rely on RGBD sensors have been introduced-among them [22] which is subject to object classification, pose estimation and reconstruction. Similar to us the training data set is composed of depth and image intensity cues and the object classes are detected using a modified Hough transform.…”

Real-time 3D segmentation of cluttered scenes for robot grasping

“…Although robust grasp planning and execution approaches are available, they typically require precise shape and pose information to select a grasp in an offline optimization process [1], [2]. In order to acquire the necessary shape and pose information, traditional approaches typically employ a-priori knowledge about object models [1], [3], [4], [5], which is used for object recognition and the subsequent planning process. However, this approach restricts grasping to objects whose models, i.e.…”