Automated Design of Salient Object Detection Algorithms with Brain Programming

Abstract

Despite recent improvements in computer vision, artificial visual systems’ design is still daunting since an explanation of visual computing algorithms remains elusive. Salient object detection is one problem that is still open due to the difficulty of understanding the brain’s inner workings. Progress in this research area follows the traditional path of hand-made designs using neuroscience knowledge or, more recently, deep learning, a particular branch of machine learning. Recently, a different approach based on genetic programming appeared to enhance handcrafted techniques following two different strategies. The first method follows the idea of combining previous hand-made methods through genetic programming and fuzzy logic. The second approach improves the inner computational structures of basic hand-made models through artificial evolution. This research proposes expanding the artificial dorsal stream using a recent proposal based on symbolic learning to solve salient object detection problems following the second technique. This approach applies the fusion of visual saliency and image segmentation algorithms as a template. The proposed methodology discovers several critical structures in the template through artificial evolution. We present results on a benchmark designed by experts with outstanding results in an extensive comparison with the state of the art, including classical methods and deep learning approaches to highlight the importance of symbolic learning in visual saliency.

Brain Programming and Its Resilience Using a Real-World Database of a Snowy Plover Shorebird

“…The strategy follows a goal-oriented framework where we study learning as a symbolic optimization process where an individual consists of a template describing the VA model while discovering critical parts of the algorithm through artificial evolution. This study focuses on the mathematical notation to formulate the SOD problem from the viewpoint of robustness and leaves the full explanation of the algorithm to consult in a previous document [8]. The input convolutional layer and the first four stages are adopted from [11], the transfer Residual Neural Network (ResNet-34), while the residual refine module consists of filters followed by a batch normalization and a Rectified Linear Unit (ReLU) activation function [12].…”

Digital Visual Design Reengineering and Application Based on K-means Clustering Algorithm

“…Literature [11] proposes a color system extraction method based on K-means clustering algorithm, and constructs a color database of aesthetic artifacts; literature [12] adopts an image processing method, extracts the mural color system from the color three-channel, and puts forward a gene extraction method based on machine learning algorithms for the digital visual design recreation; literature [13] studies the color particles of the mural paintings, and through the particle swarm optimization algorithm Improved K-means clustering analysis of color features; Literature [14] uses deep learning methods to extract and learn graphic shapes, and constructs a digital reengineering model of multi-dimensional linear structure shapes; Literature [15] improves the K-means clustering method to digitally extract texture representations by using the peak density strategy, and at the same time, uses a self-coder neural network to construct the artifacts graphic texture expression model ; Literature [16] combines the gray wolf optimization algorithm, K-means algorithm and convolutional neural network method table mural texture aesthetics for feature extraction and reengineering representation; Literature [17] combines the color characteristics of aesthetic artifacts and line characteristics, to build a digital visual design reengineering evaluation system. For the analysis of the above literature, the existing graphic visual gene extraction methods have the following defects [18]: 1) digital visual design reengineering feature selection is not standard enough [19]; 2) gene extraction methods lack of generalization [20]; 3) gene extraction methods are not efficient enough.…”