A method and device for scene and target recognition based on multi-task learning

Abstract

The invention relates to a method and device for scene and target recognition based on multi-task learning. The method includes: collecting pictures containing different scenes and targets as image sample data; manually labeling the image sample data to obtain target category labels and Scene category label; build a multi-layer convolutional neural network model and initialize the network; use image sample data and corresponding target category labels to pre-train the constructed model until convergence, and obtain a target recognition model; based on multi-task learning technology , adding a network branch to a specific layer of the target recognition model and initializing it randomly to obtain a multi-task network; using the image sample data and the corresponding scene category labels and target category labels, the multi-task network is retrained until convergence, and the multi-task network is obtained. Learning model; input new image data to the multi-task learning model to obtain the classification results of image scene and target recognition. It improves single-task recognition accuracy.

Description

technical field [0001] The invention relates to the combination of the field of vision, image recognition and deep learning, in particular to a method and device for scene and target recognition based on multi-task learning. Background technique [0002] With the rise of deep learning, more and more technologies use deep learning to realize image recognition of pictures or video streams. Compared with traditional methods, deep learning avoids the complexity of manual parameter adjustment and manual feature selection. By building a deep network model, multi-layer analysis and abstract feature extraction are performed on data, which has high accuracy, high reliability, and high Adaptive features. Common image recognition applications cover action recognition, face recognition, object recognition, scene recognition, etc. Among them, object recognition and scene recognition, as the basis of image retrieval, image classification, scene understanding, and environment perception,...

AI Technical Summary

Problems solved by technology

However, since the scene is related to multi-layer information such as the target, the background environment, and the spatial layout, the scene features of the same category have strong variability, and the scene features of different categories may have similar points overlapping each other, which brings great difficulties for accurate recognition. great challenge

At present, the overall scene recognition based on cutting-edge deep learning technology can only achieve a top-1 accuracy rate of about 50%, and it is difficult to play an effective role in practical applications.

Images