In order to enable the industrial robots to recognize the specific targets quickly and accurately on the
assembly line, an object recognition method driven by visual selective attention mechanism is proposed.
With mass training data and a machine learning model containing a number of hidden layers, deep
learning can learn more useful features, and thus ultimately improve the classification or the prediction
accuracy. The main idea of this method is as follows: for all part images, the visual attention mechanism
is used to choose salient regions in an image, achieving the goal of target segmentation. Then an
image recognition method based on deep learning is applied to recognize the chosen salient regions.
Experimental results show the effectiveness of the proposed method and the cognitive rationality.