A small target detection method for road pedestrians based on clustering idea
A technology for small target detection and pedestrian detection. It is applied in the fields of image processing, target detection and deep learning. It can solve the problems of low detection efficiency and achieve the effect of enhancing detection ability, fast detection speed and wide application range.
Active Publication Date: 2022-07-26
UNIV OF ELECTRONICS SCI & TECH OF CHINA
View PDF9 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0004] The purpose of the present invention is: the present invention provides a method for detecting small pedestrians on the road based on the idea of clustering, which overcomes the problem of low detection efficiency when the existing method divides a large image into small images to detect small pedestrians, and improves the detection efficiency of large pedestrians. Ability and Efficiency of Small Target Pedestrian Detection in Images
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment 1
[0059] like Figure 1-15 As shown, a small target detection method for road pedestrians based on clustering idea, the implementation process is as follows figure 1 shown, including the following steps:
[0060] Step 1: Build pedestrian clustering labels and train a clustering model based on a general deep learning target detection framework;
[0061] Further, the specific implementation steps of the step 1 are as follows:
[0062] Step 1.1: Perform clustering transformation on all labels of the existing pedestrian detection dataset CityPersons to obtain pedestrian clustering labels. First, obtain all the pedestrian bounding box annotation information in each image Among them, i represents the index of each object in the image, and g i ={x 1i , y 1i , x 2i , y 2i }, (x 1 , y 1 ) and (x 2 , y 2 ) respectively represent the abscissa and ordinate of the upper left corner and the lower right corner of the labeling frame of the object; secondly, calculate the center poin...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention discloses a small target detection method for road pedestrians based on a clustering idea, which relates to the technical fields of image processing, target detection and deep learning. Clustering model; S2: Input the image to be detected; S3: Use the clustering model to extract the pedestrian clustering area in the image to be detected; S4: Preprocess all the pedestrian clustering areas, and adjust the size of the clustering area to make it match the Match the input size of the pedestrian detection model; S5: Train a pedestrian detection model based on the general target detection model, perform fine detection on the adjusted pedestrian clustering areas, and obtain pedestrian detection results in all clustered areas; S6: Use non-polarity The large value suppression processes all pedestrian detection results, and maps the location information of pedestrian detection results in all clustered areas to the original image to be detected; S7: Output all pedestrian detection results in the image to be detected.
Description
technical field [0001] The invention relates to the technical fields of image processing, target detection and deep learning, in particular to a small target detection method for road pedestrians based on clustering ideas. Background technique [0002] The problem of pedestrian small target detection has always been a difficult problem in pedestrian detection tasks, mainly because the pedestrian small target is blurred in the image, has low resolution and carries little information, which leads to weak feature expression ability. In the process of feature extraction, it can be extracted. There are very few features, so the detection accuracy of small pedestrian targets is usually only half of that of large pedestrian targets. General small target detection schemes mainly include: using image pyramids and multi-scale sliding windows, such as MTCNN, FPN and Feature-FusedSSD, etc.; using data enhancement methods, such as oversampling and copy-pasting small targets; using differ...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Patents(China)
IPC IPC(8): G06V40/10G06V10/762G06V10/82G06K9/62G06N3/04G06N3/08
CPCG06N3/04G06N3/08G06V20/53G06V2201/07G06F18/23
Inventor 袁国慧叶涛王卓然
Owner UNIV OF ELECTRONICS SCI & TECH OF CHINA