Small target object detection method
An object detection, small target technology, applied in the computer field, can solve the problems of low accuracy, easy to see dazzling eyes, high labor cost, and achieve the effect of accurate prediction results
Pending Publication Date: 2021-03-30
CHINA TEXTILE ENG SOC +2
View PDF5 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
The whole process is heavy, boring and repetitive, and because many of the colonies are very small, it is very easy for the human eye to be dazzled during the counting process, and it is necessary to concentrate on watching, and it is easy to fall into fatigue, which leads to inefficiency. low, high labor cost
[0003] In order to improve efficiency and reduce labor costs, the current detection of small target objects mainly uses the convolutional neural network model to perform convolution and pooling operations on the original image to obtain feature images of different sizes and predict the feature images of the last layer. Feature images of other layers, less accurate
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 Construction
[0021] The present invention will be specifically introduced below in conjunction with the accompanying drawings and specific embodiments.
[0022] Such as figure 1 As shown, the small target object detection method provided by the embodiment of the present invention includes the following steps:
[0023] S101, using the convolutional neural network model to continuously down-sample the original image where the small target object is located, to obtain multiple feature images;
[0024] S102, the convolutional neural network model sorts a plurality of feature images according to the time sequence of downsampling, to obtain a feature image group;
[0025] S103, the convolutional neural network model selects the last n feature images in the feature image group, and records the n feature images as C in sequence according to the order of size 1 、C 2 、C 3 …C n ;
[0026] S104, using the convolutional neural network model, the feature image C n Perform 1*1 convolution operatio...
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 object detection method, and relates to the technical field of computers. A feature image collected in each convolution and downsampling processis used and eachfeature and a feature output by the next downsampling process are superposed to predict a small target object by using a large-size feature image with little lost information, A small-size feature image with more abstract semantic information obtained through multiple convolution is used for predicting a large target object, and the accuracy of a prediction result is improved.
Description
technical field [0001] The invention relates to the technical field of computers, in particular to a method for detecting small target objects. Background technique [0002] For the detection of small target objects such as colonies, the traditional solution is to first cultivate colonies on a petri dish, then use human eyes to observe and mark and count on the petri dish with a black marker. The whole process is heavy, boring and repetitive, and because many of the colonies are very small, it is very easy for the human eye to be dazzled during the counting process, and it is necessary to concentrate on watching, and it is easy to fall into fatigue, which leads to inefficiency. Low, high labor costs. [0003] In order to improve efficiency and reduce labor costs, the current detection of small target objects mainly uses the convolutional neural network model to perform convolution and pooling operations on the original image to obtain feature images of different sizes and p...
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 Applications(China)
IPC IPC(8): G06T7/00G06K9/46G06N3/04G06N3/08
CPCG06T7/0012G06N3/08G06T2207/10056G06T2207/30004G06V10/40G06N3/045
Inventor 伏广伟张珍竹崔绮嫦王斌罗桂莲郑少锋何南坚李伟才黄慧宇王文余娟毕兴忠王军
Owner CHINA TEXTILE ENG SOC