TARGET DETECTION METHOD AND APPARATUS BASED ON ATTENTION MECHANISM, AND COMPUTER DEVICE

摘要

A target detection method and apparatus based on an attention mechanism, and a computer device. The method comprises: receiving an image to be detected input by a user; inputting said image into a convolutional neural network model, and extracting a multilayer feature map of said image; weighting the multilayer feature map according to an attention mechanism, so as to obtain a weighted feature map; generating a feature pyramid of said image according to the multilayer feature map; respectively fusing the weighted feature map with a feature map of each layer in the feature pyramid, so as to obtain a fused feature pyramid; acquiring, from the fused feature pyramid, a feature map that matches a target image; and according to a target detection model, performing target detection on the feature map that matches the target image, so as to obtain the target image. According to the method, on the basis of neural network technology in artificial intelligence, an attention mechanism is introduced to fuse features of a convolutional output layer, thereby greatly improving the accuracy when performing different target detection tasks.