The representation, recognition, and locating of 3-d objects
International Journal of Robotics Research
Algorithms for clustering data
Algorithms for clustering data
Evidence-Based Recognition of 3-D Objects
IEEE Transactions on Pattern Analysis and Machine Intelligence
Fundamentals of digital image processing
Fundamentals of digital image processing
IEEE Transactions on Pattern Analysis and Machine Intelligence
Self-organization and associative memory: 3rd edition
Self-organization and associative memory: 3rd edition
ALVINN: an autonomous land vehicle in a neural network
Advances in neural information processing systems 1
Learning internal representations by error propagation
Parallel distributed processing: explorations in the microstructure of cognition, vol. 1
Invariant Descriptors for 3D Object Recognition and Pose
IEEE Transactions on Pattern Analysis and Machine Intelligence - Special issue on interpretation of 3-D scenes—part I
Structural Indexing: Efficient 3-D Object Recognition
IEEE Transactions on Pattern Analysis and Machine Intelligence - Special issue on interpretation of 3-D scenes—part II
Why progress in machine vision is so slow
Pattern Recognition Letters
Geometric invariants and object recognition
International Journal of Computer Vision
Perceptual Organization and Visual Recognition
Perceptual Organization and Visual Recognition
Machine Learning
Genetic algorithms for object recognition in a complex scene
ICIP '95 Proceedings of the 1995 International Conference on Image Processing (Vol.2)-Volume 2 - Volume 2
Minimizing Binding Errors Using Learned Conjunctive Features
Neural Computation
Minimizing Binding Errors Using Learned Conjunctive Features
Neural Computation
Temporal context as cortical spatial codes
IJCNN'09 Proceedings of the 2009 international joint conference on Neural Networks
Hi-index | 0.00 |
This paper presents a framework called Cresceptron for view-basedlearning, recognition and segmentation. Specifically, it recognizesand segments image patterns that are similar to those learned, usinga stochastic distortion model and view-based interpolation, allowingother view points that are moderately different from those used inlearning. The learning phase is interactive. The user trains thesystem using a collection of training images. For each trainingimage, the user manually draws a polygon outlining the region ofinterest and types in the label of its class. Then, from thedirectional edges of each of the segmented regions, the Cresceptronuses a hierarchical self-organization scheme to grow a sparselyconnected network automatically, adaptively and incrementally duringthe learning phase. At each level, the system detects new imagestructures that need to be learned and assigns a new neural plane foreach new feature. The network grows by creating new nodes andconnections which memorize the new image structures and their contextas they are detected. Thus, the structure of the network is afunction of the training exemplars. The Cresceptron incorporates bothindividual learning and class learning; with the former, eachtraining example is treated as a different individual while with thelatter, each example is a sample of a class. In the performancephase, segmentation and recognition are tightly coupled. Noforeground extraction is necessary, which is achieved by backtrackingthe response of the network down the hierarchy to the image partscontributing to recognition. Several stochastic shape distortionmodels are analyzed to show why multilevel matching such as that inthe Cresceptron can deal with more general stochastic distortionsthat a single-level matching scheme cannot. The system isdemonstrated using images from broadcast television and other videosegments to learn faces and other objects, and then later to locateand to recognize similar, but possibly distorted, views of the sameobjects.