GRUFF-3: Generalizing the domain of a function-based recognition system

doi:10.1016/0031-3203(94)90091-4

Pattern Recognition

Volume 27, Issue 12, December 1994, Pages 1743-1766

https://doi.org/10.1016/0031-3203(94)90091-4 Get rights and content

Abstract

Representation systems which support “generic” object recognition offer promising advantages over current model-based vision. Systems applying function-based reasoning are one such approach. In this approach, specific geometric or structural models are disregarded, in favor of analyzing the shape to determine functional requirements for category membership. This paper presents an explanation of the ideas behind function-based modeling and a description of the extensions made to create the Generic Representation Using Form and Function-3 (GRUFF-3) system. This system analyzes the 3D shape of an object and classifies the object according to its possible function as some (sub) category of the superordinate category dishes. The initial GRUFF system implementation was restricted to the furniture domain and required five knowledge primitives (clearance, relative orientation, proximity, dimensions and stability) to realize the functional requirements of the categories represented. The important contribution of our current work is that a significantly larger domain of objects can now be recognized with the addition of just one new knowledge primitive, enclosure. An evaluation of the performance of the system is presented for a database of over 200 3D shapes.

References (23)

E. Rosch et al.
Basic objects in natural categories
Cognit. Psych.
(1976)
J.H. Connell et al.
Generating and generalizing models of visual objects
Artif. Intell.
(1987)
L. Stark et al.
Function-based recognition for multiple object categories
Image Understanding
(1994)
J. Aloimonos
Purposive and qualitative action vision
K. Ikeuchi et al.
Task-oriented vision
M. Minsky
The Society of Mind
(1985)
T.O. Binford
Survey of model-based image analysis systems
Int. J. Robotics Res.
(1982)
P. Winston et al.
Learning physical descriptions from functional definitions, examples, and precedents
M. Brady et al.
The mechanics mate
M. Dimanzo et al.
FUR: understanding functional reasoning
Int. J. Intell. Syst.
(1989)

M. Brand

Vision systems that see in terms of function

Working Notes on Reasoning About Function

(1993)

Cited by (21)

Learning function-based object classification from 3D imagery
2008, Computer Vision and Image Understanding
Citation Excerpt :
The possible advantages of functional approaches for generic classification were recognized in several relatively early works, such as [11] and [51]. Following these concepts, several systems for object classification were built (see [1,11,41,43]). However, little experimental work has been done to test these concepts.
We propose a novel scheme for using supervised learning for function-based classification of objects in 3D images. During the learning process, a generic multi-level hierarchical description of object classes is constructed. The object classes are described in terms of functional components. The multi-level hierarchy is designed and constructed using a large set of signature-based reasoning and grading mechanisms. This set employs likelihood functions that are built as radial-based functions from the histograms of the object instances. During classification, a probabilistic matching measure is used to search through a finite graph to find the best assignment of geometric parts to the functional structures of each class. An object is assigned to the class that provides the highest matching value. Reuse of functional primitives in different classes enables easy expansion to new categories. We tested the proposed scheme on a database of about 1000 different 3D objects. The proposed scheme achieved high classification accuracy while using small training sets.
Efficient search and verification for function based classification from real range images
2007, Computer Vision and Image Understanding
Citation Excerpt :
The authors in [16] present the FUR (FUnctional Reasoning) project, a functional reasoning and shape–function integration system, in which several functions (such as support, grasp, enter, and hang) are presented, and the use of functional expert concepts for identification of functional primitives is discussed. An impressive number of good results in the function-based classification field were demonstrated with the GRUFF, OMLET, and OPUS systems [22,48,49,51,56]. GRUFF, which employs generic representations using form and function, was extensively used on more than two hundred synthetic models of five categories [51].
In this work we propose a probabilistic model for generic object classification from raw range images. Our approach supports a validation process in which classes are verified using a functional class graph in which functional parts and their realization hypotheses are explored. The validation tree is efficiently searched. Some functional requirements are validated in a final procedure for more efficient separation of objects from non-objects. The search employs a knowledge repository mechanism that monotonically adds knowledge during the search and speeds up the classification process. Finally, we describe our implementation and present results of experiments on a database that comprises about 150 real raw range images of object instances from 10 classes.
More than just shape: A representation for functionality
1998, Artificial Intelligence in Engineering
This paper presents a representation for objects based not only on shape and material properties but also on the role that the object plays in the scene. Functionality is introduced as a representation mapping the intrinsic properties of an object in the perceptual and interactive space of an agent and is characterized as an extension of purposive and active vision, active perception and task-oriented approaches. The developed robotic system extracts functional features employed in interactions for piercing and chopping. The representation is presented by means of histograms and Force–Shape maps. Histograms are provided as means of representing the success in accomplishing the various experiments utilizing different tools and materials. Force–Shape maps are employed to express the dynamics of functional interactions. The approach described in this paper has two goals: a description of the components constituting a representation of functionality and a methodology for a robotic agent to extract a representation.
Functional features for chopping extracted from observations and interactions
1998, Image and Vision Computing
This article presents the developments of our research for the identification and extraction of functional features through the interactions and observations performed by a robotic system. The ability to identify and extract functional features and the recovery of a representation for a functionality, are fundamental to robotic agents interacting with an unknown and unstructured environment. A functional interpretation characterizes objects with respect to properties which make them useful for specific interactions. A case study involving the functionality of chopping is discussed. Geometric properties of the tools, expressed using superellipses, forces and task description of the interaction are combined into success-histograms and force-shape maps. These descriptions characterize the functional interaction and combine data from different sensory modalities identifying functional features.
Understanding object motion
1998, Image and Vision Computing
Many types of common objects, such as tools and vehicles, usually move in simple ways when they are wielded or driven: the natural axes of the object tend to remain aligned with the local trihedron defined by the object's trajectory. Based on this observation we use a model called Frenet-Serret motion which corresponds to the motion of a moving trihedron along a space curve. Knowing how the Frenet-Serret frame is changing relative to the observer gives us essential information for understanding the object's motion. This is illustrated here for four examples, involving tools (a wrench and a saw) and vehicles (an accelerating van, a turning taxi).
Function from visual analysis and physical interaction: A methodology for recognition of generic classes of objects
1998, Image and Vision Computing
This paper presents an overview of the GRUFF-I (Generic Recognition Using Form, Function and Interaction) system, a nonpart-based approach to generic object recognition which reasons about and generates plans for interaction with three-dimensional (3D) shapes from the categories furniture and dishes. The system operates as follows. A researcher selects an object and places it in an observation area. An initial intensity and range image are acquired. These are input to a three-stage recognition system. The first stage builds a 3D model. The second stage receives as input a 3D model and considers the shape-suggested functionality of this shape by applying concepts of physics and causation (e.g. to infer stability) to label the object's potential functionality. The third stage uses this labeling to instantiate a plan for interaction to confirm the object's functional use in a task by incorporating feedback from both visual and robotic sensors. Results of this work are presented for eighteen chair-like and cup-like objects. Major conclusions from this work include: (1) metrically accurate representations of the world can be built and used for higher level reasoning; (2) shape-based reasoning prior to interaction-based reasoning provides an efficient methodology for object recognition, in terms of the judicious use of system resources; and (3) interaction-based reasoning can be used to confirm the functionality of a categorized object without explicitly determining the object's material composition.

View all citing articles on Scopus

^☆: This research was supported by AFOSR grant F4962092-J-0223, NSF grant IRI-9120895, and a NASA Florida Space Grant Consortium graduate fellowship.

View full text

GRUFF-3: Generalizing the domain of a function-based recognition system☆

Abstract

Cognit. Psych.

Artif. Intell.

Image Understanding

Purposive and qualitative action vision

Task-oriented vision

The Society of Mind

Survey of model-based image analysis systems

Int. J. Robotics Res.

Learning physical descriptions from functional definitions, examples, and precedents

The mechanics mate

FUR: understanding functional reasoning

Int. J. Intell. Syst.

Vision systems that see in terms of function

Working Notes on Reasoning About Function