Image characteristic combination determination method based on deep neural network

The invention relates to an image characteristic combination determination method based on a deep neural network. The method comprises the following steps of S1, through a first deep neural network, carrying out characteristic extraction on an input image and generating a first result, and acquiring a first predicted value; S2, comparing the first predicted value with a first preset threshold anda second preset threshold; S3, if the first predicted value is between the first threshold and the second threshold, through a second deep neural network, carrying out characteristic extraction on theinput image and generating a second result, and acquiring a second predicted value; and S4, through the first predicted value and the second predicted value, matching the first result and the secondresult, and if matching is successful, generating the bounding box of the first result. Through combination determination, the correlation degree of the first result and the second result is effectively determined, the position