Processing occlusions using elastic-net hierarchical MAX model of the visual cortex

Alameer, A; Degenaar, P; Nazarpour, K

Abstract

Humans can recognise objects under partial occlusion. Machine-based approaches cannot reliably recognise objects and scenes in the presence of occlusion. This paper investigates the use of the elastic net hierarchical MAX (En-HMAX) model to handle occlusions. Our experiments show that the En-HMAX model achieves an accuracy of ~70%, when ~50% artificial occlusions are applied to the centre of the visual object-field. Furthermore, when the same percentage of occlusion is applied to the peripheral, the model reports higher accuracies. A similar degree of robustness has been observed when recognising scenes. The results suggest that cortex-like models, such as the En-HMAX are reliable for solving the occlusion challenge.

Citation

Alameer, A., Degenaar, P., & Nazarpour, K. (2017, July). Processing occlusions using elastic-net hierarchical MAX model of the visual cortex. Presented at 2017 IEEE International Conference on INnovations in Intelligent SysTems and Applications (INISTA), Gdynia, Poland

• Presentation Conference Type: Other
• Conference Name: 2017 IEEE International Conference on INnovations in Intelligent SysTems and Applications (INISTA)
• Conference Location: Gdynia, Poland
• Start Date: Jul 3, 2017
• End Date: Jul 5, 2017
• Online Publication Date: Aug 8, 2017
• Publication Date: Aug 8, 2017
• Deposit Date: Jun 9, 2022
• DOI: https://doi.org/10.1109/INISTA.2017.8001150
• Publisher URL: https://doi.org/10.1109/INISTA.2017.8001150
• Additional Information: Event Type : Conference