Related Experiment Video
Updated: May 15, 2026

End-To-End Deep Neural Network for Salient Object Detection in Complex Environments
Published on: December 15, 2023
Perceptual organization and artificial attention for visual landmarks detection.
Esther Antúnez1, Antonio J Palomino, Rebeca Marfil
1Grupo ISIS, Departamento Tecnología Electrónica, Universidad de Málaga, Málaga, Spain. eantunez@uma.es
This study introduces an artificial attention model for mobile robot navigation, enhancing computational efficiency by focusing on visual landmarks. It employs bottom-up and top-down saliency maps for object-level attention and landmark re-detection.
Area of Science:
- Computer Vision
- Robotics
- Artificial Intelligence
Background:
- Biological attention mechanisms optimize information processing in vision.
- Artificial attention models aim to improve computational efficiency in vision systems.
- Mobile robotics navigation requires robust visual processing and localization.
Purpose of the Study:
- To propose an artificial attention model for mobile robot navigation.
- To implement object-level attention using bottom-up and top-down saliency maps.
- To enable efficient landmark detection and re-detection for robot localization.
Main Methods:
- Bottom-up attention implemented via a hierarchical process using a Combinatorial Pyramid for perceptual grouping of image regions and edges.
- Top-down attention utilizing combinatorial submaps for error-tolerant landmark re-detection.
- Integration of saliency map estimation for guiding attention.
Main Results:
- Demonstrated a hierarchical process for perceptual grouping and saliency map generation.
- Developed an error-tolerant submap isomorphism procedure for landmark re-detection.
- Successfully applied object-level attention to mobile robot navigation challenges.
Conclusions:
- The proposed artificial attention model enhances computational resource optimization in mobile robot navigation.
- The integration of bottom-up and top-down attention mechanisms improves landmark recognition and re-detection.
- This approach offers a promising direction for developing more efficient and capable autonomous robots.
More Related Videos
Related Concept Videos
Gestalt Principles of Perception
Perceptual Constancy
Size constancy is the recognition that an object remains the same size, even when its image on the retina changes. For instance, a bus is perceived to be large enough to carry people, even if it looks tiny from...
Depth Perception and Spatial Vision
Perception
Bottom-up processing begins at the sensory level, where receptors detect external environmental stimuli. These could include the tactile sensation of...
Parallel Processing
Visual System
Once through the pupil, the light passes through the lens, a...

