Research On The Application Of Deep Learning In The Recognition Of Cognitive Salient Features Of Car Front Face Image

In the current society,cars have become an essential means of transportation in modern society.The front face design of a car is one of the critical factors for consumers to have a good emotional experience and to make purchase decisions.It is also a key design element for consumers’ perceptual image cognition and emotional experiences,which has attracted increasing attention in automotive design.With the rapid development of deep learning,it provides powerful technical support for the emotional design of automotive front faces.Therefore,this article applies deep learning technology in the recognition and characterization of automotive front image saliency features.Using image saliency features as a basis,it achieves the intelligent design of the emotional scheme for the front face of automobiles.It is aimed to generate a more emotionally appealing design for the front face of cars.The research process mainly consists of three aspects: dataset construction,perceptual image recognition classification model and image saliency feature recognition model construction,and intelligent generation of automotive frontal face emotional design.Firstly,a car exterior network comment text dataset and a representative car frontal face dataset were constructed through data collection.A subset of samples from the image components were selected to construct the representative frontal face dataset.The word2 vec model was used to select representative image vocabulary to form an image word space,combined with the importance experiment to obtain core image vocabulary.The frontal face image classification dataset was constructed through manual annotation with the expert evaluation method.Secondly,a perceptual image recognition classification model for automotive frontal faces was built using Res Net.Based on the weighted feature heat map of Grad-CAM,the visualized interpretation of classification basis was achieved,and the conspicuous image features of automotive frontal faces were identified,providing data support for emotional design of automotive frontal faces.Finally,based on the idea of self-learning of deep learning image saliency features,automotive frontal face generation models such as DCGAN,c DCGAN,and Cycle GAN were constructed.These models were used to intelligently generate automotive frontal face emotional design schemes with specific target image or continuing product family features.The accuracy of the target image was verified based on the image recognition classification model,demonstrating the effectiveness of the intelligent emotional design generation model.This study is based on deep learning to recognize the image saliency features of automotive frontal faces,and intelligently generate automotive frontal face images with specific target images.It improves the research and development efficiency of automotive frontal face design,shortens the product iteration cycle,and has broad application prospects in future automotive design and development.