Towards automated creation of image interpretation systems
- Ilya Levner
- Vadim Bulitko
- Lihong Li
- Greg Lee
- Russ Greiner, Dept of Computing Science; PI of AICML
Automated image interpretation is an important task in numerousapplications ranging from security systems to natural resourceinventorization based on remote-sensing. Recently, a second generationof adaptive machine-learned image interpretation systems have shownexpert-level performance in several challenging domains. While demonstratingan unprecedented improvement over hand-engineered and firstgeneration machine-learned systems in terms of cross-domain portability,design-cycle time, and robustness, such systems are still severely limited.This paper inspects the anatomy of the state-of-the-art Multi resolutionAdaptive Object Recognition framework (MR ADORE) and presents extensionsthat aim at removing the last vestiges of human intervention stillpresent in the original design of ADORE. More specifically, feature selectionis still a task performed by human domain experts and represents amajor stumbling block in the creation process of fully autonomous imageinterpretation systems. This paper focuses on minimizing such need forhuman engineering. After discussing experimental results, showing theperformance of the framework extensions in the domain of forestry, thepaper concludes by outlining autonomous feature extraction methodsthat may completely remove the need for human expertise in the featureselection process.
Citation
I. Levner, V. Bulitko, L. Li, G. Lee, R. Greiner. "Towards automated creation of image interpretation systems". Australian Joint Conference on Artificial Intelligence, pp 653-665, July 2003.Keywords: | vision, machine learning, reinforcement learning, MrAdore |
Category: | In Conference |
BibTeX
@incollection{Levner+al:AustralianAI03, author = {Ilya Levner and Vadim Bulitko and Lihong Li and Greg Lee and Russ Greiner}, title = {Towards automated creation of image interpretation systems}, Pages = {653-665}, booktitle = {Australian Joint Conference on Artificial Intelligence}, year = 2003, }Last Updated: June 05, 2007
Submitted by Staurt H. Johnson