Method for detecting image target in computer vision

The invention discloses a method for detecting an image target in computer vision, which comprises the following steps: 1, setting a screening candidate box threshold value Sc, and setting the confidence coefficient of a candidate box with lower confidence coefficient to be zero; step 2, determining a threshold Nt for screening candidate frames with higher scores, dividing the candidate frames into two parts according to IOU scores, retaining confidence scores of the candidate frames smaller than Nt, and recombining the candidate frames larger than Nt; step 3, dividing a threshold value range in a range of [Nt, 1] into three sections of [Nt, N1), [N1, N2) and [N2, 1], respectively setting different attenuation factors according to a formula (3), inhibiting the confidence coefficient of the candidate frame of which the IOU is greater than Nt, and re-arranging the scores of the inhibited IOU from large to small; and 4, repeating the steps 2 and 3 until all candidate boxes are screened. According to the technica