CSAIL Research Abstract. Probabilistic Geometric Grammars for Object Recognition Meg Aycinena, Leslie Pack Kaelbling & Tomas Lozano-Perez Introduction We are researching a generative parts-based three-dimensional representation and recognition framework for classes of objects. The framework uses probabilistic grammars to represent object classes recursively in terms of their parts, thus exploiting the hierarchical and substitutive structure inherent to many types of objects.
It models the 3D geometric characteristics of object parts using multivariate conditional Gaussians over dimensions, position, and rotation. We develop algorithms for learning geometric models and rule probabilities given parsed 3D examples and a fixed grammar. Motivation This work is novel in that it combines several approaches to the task of object classification and recognition: the use of three-dimensional models, a parts-based approach, and the use of probabilistic grammars to capture structural variability. Approach Assumptions References. RGB-D Object Recognition and Detection. In this project we address joint object category, instance, and pose recognition in the context of rapid advances of RGB-D cameras that combine both visual and 3D shape information. RGB-D cameras are the underlying sensing technology behind Microsoft Kinect. The focus of this project is on detection and classification of objects in indoor scenes, such as in domestic environments. Overview Personal robotics is an exciting research frontier with a range of potential applications including domestic housekeeping, caring of the sick and the elderly, and office assistants for boosting work productivity.
The ability to detect and identify objects in the environment is important if robots are to safely and effectively perform useful tasks in unstructured, dynamic environments such as our homes, offices and hospitals. Detection-based Object Labeling in 3D Scenes In this work we propose a view-based approach for labeling objects in 3D scenes reconstructed from RGB-D (color+depth) videos.
PlayMate - Activities. 3D Object Recognition. This dataset contains multi-view images for 300 3D objects that are rigid and sufficiently textured. Each 3D object has ~250 images (1280 by 720 pixels) captured from different viewpoints with a clean background, using a turntable and a webcam (with a fixed focal length 1000 pixels).
The dataset can be used for research on 3D reconstruction and object recognition. Format The data for each 3D object is provided in a separate folder, which contains:1) a "db_img" subfolder for the multi-view images of the object;2) a "list_db_img.txt" file that lists the filenames of the images;3) a "model.nvm" file containing an example 3D point cloud model reconstructed from the images using VisualSFM toolkit ( The model file (after copied to the image subfolder) can be opened with VisualSFM to show a possible 3D representation of the object.
Download Reference Contact If you have questions about the dataset, please contact Qiang Hao (email: haoq@live.com). Addressing Feature Ambiguity for Point-Based 3D Object Recognition. Addressing Feature Ambiguity for Point-Based 3D Object Recognition People Edward HsiaoAlvaro Collet RomeaMartial Hebert Description In many of the current point-based 3D object recognition systems, specific point-to-point correspondences from 3D model to 2D image are obtained initially by matching discriminative features. Much of the recent research in local image features has been to design descriptors that are as discriminative and robust as possible to obtain point-to-point matches.
We propose to maintain feature ambiguity by quantizing the features on a model. Another issue that arises in the real world is that objects in unstructured environments can appear in any orientation and position, often significantly different from the images used to train the model. Existing 3D recognition datasets have a large number of objects, but are mainly comprised of objects on monotone backgrounds.
Dataset References [1] Edward Hsiao, Alvaro Collet and Martial Hebert. Funding Copyright notice.