The representation, recognition, and locating of 3-d objects
International Journal of Robotics Research
Incremental reconstruction of 3D scenes from multiple, complex images
Artificial Intelligence
Recognition of occluded objects: a cluster-structure algorithm
Pattern Recognition
Transformational invariance: a primer
Image and Vision Computing
Retrieving Multispectral Satellite Images Using Physics-Based Invariant Representations
IEEE Transactions on Pattern Analysis and Machine Intelligence
Continuous case-based reasoning
Artificial Intelligence
Neural Network Perception for Mobile Robot Guidance
Neural Network Perception for Mobile Robot Guidance
Genetic Learning for Adaptive Image Segmentation
Genetic Learning for Adaptive Image Segmentation
Robot Learning
Digital Image Processing
Closed-Loop Object Recognition Using Reinforcement Learning
CVPR '96 Proceedings of the 1996 Conference on Computer Vision and Pattern Recognition (CVPR '96)
Reading Checks with Multilayer Graph Transformer Networks
ICASSP '97 Proceedings of the 1997 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP '97) -Volume 1 - Volume 1
Some Extensions of the K-Means Algorithm for Image Segmentation and Pattern Classification
Some Extensions of the K-Means Algorithm for Image Segmentation and Pattern Classification
Exploiting temporal and spatial constraints in traffic sign detection from a moving vehicle
Machine Vision and Applications
Hi-index | 0.00 |
Current machine perception techniques that typically use segmentationfollowed by object recognition lack the required robustness to copewith the large variety of situations encountered in real-worldnavigation. Many existing techniques are brittle in the sense thateven minor changes in the expected task environment (e.g., differentlighting conditions, geometrical distortion, etc.) can severelydegrade the performance of the system or even make it failcompletely. In this paper we present a system that achieves robustperformance by using local reinforcement learning to induce a highlyadaptive mapping from input images to segmentation strategies forsuccessful recognition. This is accomplished by using the confidencelevel of model matching as reinforcement to drive learning. Localreinforcement learning gives rises to better improvement inrecognition performance. The system is verified through experimentson a large set of real images of traffic signs.