Fine-grained image recognition method and device based on weak supervised learning, and readable medium

Abstract

The invention discloses a fine-grained image recognition method and device based on weak supervised learning and a readable medium, and the method comprises the steps: constructing a VGGRecognition model based on an attention mechanism, carrying out the two-step transfer learning training, and enabling the VGGRecognition model based on the attention mechanism to comprise a pre-trained VGG16 model, a Recognition part and an attention mechanism part, the Rescoption part comprises a fourth batch of normalization layers, a plurality of Rescoption modules, a third convolution layer and a third batch of normalization layers, each Rescoption module comprises a first convolution layer, a first batch of normalization layers, an Inception-A unit, a second convolution layer and a second batch of normalization layers which are connected based on residual errors, and the attention mechanism part comprises an attention mechanism module, a global average pooling layer, a full connection layer and a softmax layer. The two-step transfer learning training process comprises transfer learning between a source domain and a transition domain and transfer learning between the transition domain and a target domain; and acquiring a plant leaf disease degree fine-grained image, inputting the trained VGGResection model based on the attention mechanism, and outputting a classification result. And the stability and the accuracy can be improved.

Description

technical field [0001] The invention relates to the fields of deep learning and computer vision, in particular to a fine-grained image recognition method, device and readable medium based on weakly supervised learning. Background technique [0002] The purpose of fine-grained image classification is to distinguish subcategories with subtle visual differences, which is more challenging than traditional coarse-grained image classification. On the one hand, the feature differences between images are smaller, thus resulting in more subtle discriminative features. On the other hand, the training data set is limited and there are many uncertain factors in the image, such as lighting differences, background interference, etc. The key to fine-grained image classification is to obtain the most significant local difference features. According to the different supervision information requirements of the training data in the training process of the neural network, the research algorit...

AI Technical Summary

Problems solved by technology

[0003] In the past, the accurate segmentation problem required the expertise of the corresponding domain experts to judge, resulting in a significant increase in cost

Images