Target detection method and system based on deep learning

Abstract

The invention discloses a deep learning-based target detection method and system. The method comprises the following steps of obtaining a to-be-processed image; preprocessing the to-be-processed imageto obtain a first image; inputting the first image into a convolutional neural network for feature extraction to obtain a first feature map, and taking the first feature map as a first group of inputdata; executing the setting operation for n times to obtain second to n + 1th groups of input data; and predicting the types of targets contained in the to-be-processed image and the position of eachtarget according to the first to (n + 1) th groups of input data to obtain a prediction result. According to the method, the input of each convolution operation is the combination of the outputs of all previous convolution operations, so that the feature maps of different scales are fused, the parameter quantity is reduced, the complexity of the model can be controlled while the detection accuracy is improved, and the dependence on a pre-training model is eliminated. The method can be widely applied to the field of artificial intelligence.

Description

technical field [0001] The invention relates to the field of artificial intelligence, in particular to a deep learning-based target detection method and system. Background technique [0002] As a basic problem in the field of computer vision, target detection has always attracted much attention and has been widely used in unmanned driving, video surveillance, video analysis and other fields. Before the emergence of deep learning, the traditional target detection method is generally divided into three steps: first, select the area, then extract the features of the target area, and finally perform classification and regression on the prediction results. The representative methods are: background subtraction, light flow method, etc. The traditional algorithm inevitably has two shortcomings: one is that the strategy of region selection is ineffective and the time complexity is high; the other is that the robustness of manually extracted features is poor; the emergence of deep l...

AI Technical Summary

Problems solved by technology

The traditional algorithm inevitably has two shortcomings: one is that the strategy of region selection is ineffective and the time complexity is high; the other is that the robustness of manually extracted features is poor; the emergence of deep learning technology has revolutionized the mode of target detection and improved the Accuracy and Robustness of Object Detection

The object detection algorithm based on region proposal is not suitable for real-time detection due to the serious time-consuming process of extracting the region of interest

[0004] The real-time detection of the algorithm based on the frame regression is relatively better, but it also has problems such as relying too much on the pre-training model and the detection accuracy is not high enough, which needs to be improved

Images