Rapid pedestrian detection method and system
A pedestrian detection and pedestrian technology, applied in the field of target detection and computer vision, can solve the problems of incompatibility between accuracy and detection efficiency, and achieve the effect of improving pedestrian detection speed, reducing false detection rate, and improving discrimination
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0037] Such as Figure 4 As shown, the present invention discloses a fast pedestrian detection method, comprising the following steps:
[0038] Merging the MobileNet network and the RPN network to obtain a MobileNet-RPN detection model, the MobileNet-RPN detection model takes the image to be detected as input, and the predicted pedestrian frame on the image to be detected is output;
[0039] Obtaining a training data set comprising a pedestrian image marked with a real pedestrian border and a background image without pedestrians to train the MobileNet-RPN detection model to obtain a trained MobileNet-RPN detection model;
[0040] Input the image to be detected into the MobileNet-RPN detection model to obtain the predicted pedestrian frame of the image to be detected.
[0041] A fast pedestrian detection method provided by the present invention uses the MobileNet algorithm with fewer parameters to construct a lightweight feature selection network, so that the calculation amoun...
Embodiment 2
[0043] Embodiment 2 is an extended embodiment of Embodiment 1, specifically including the following:
[0044] The present invention is realized based on the deep learning open source framework Pytorch (Facebook's official deep learning framework).
[0045] S11: Using a good classification network for transfer learning and using target detection to extract image features has become the mainstream method based on deep learning in target detection. The commonly used network is VGG (Visual Geometry Group Network, Visual Geometry Group Network, Visual Geometry Group Network) and ZF-Net (deep neural network, champion of 2013 ImageNet classification task) network, however, the parameters of these two networks are extremely large, although they can extract features with strong discriminative power, they will cause the model to propagate forward The amount of calculation is too large, which makes the detection speed of the network extremely slow and cannot be applied in practice.
[0...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More 


