A small target detection method and a detection model based on a convolutional neural network

Abstract

The invention discloses a convolutional neural network-based small target detection method and a convolutional neural network-based small target detection model, and the method comprises the steps: carrying out the marking of small targets in a training set image, and building a small target data set; construction a training platform with caffe-ssd as the bottom layer; constructing a single-step detector model ELFSSD for enhancing low-layer feature fusion; adopting a model through pre-trained VGG-16 model initializing, inputting a small target data set in an lmdb format, carrying out iterativetraining; and detecting a small target in the detection set image by using the trained ELFSSD model. The method enhances the low-level characteristics, removes the high-level redundant characteristics, simplifies the detection process, improves the detection speed, accurately detects the small target in the image in real time, and solves the problem of poor detection effect of the small target inthe prior art.

Description

technical field [0001] The invention belongs to the technical field of computer vision and image processing, and relates to a small target detection method and detection model based on a convolutional neural network, which can be used for small target detection such as industrial rice quality and medical cell analysis under a microscope. Background technique [0002] At present, the rapid development of target detection technology has greatly improved the accuracy and real-time performance of target detection. From the two-step detector RCNN series model to the end-to-end network model SSD, YOLO, etc., the detection effect of the target detection model based on the convolutional neural network in the Kaggle competition and the public data set has been continuously improved, but for small target detection, existing There are obvious problems in the model and the detection effect has not been ideal. [0003] In order to improve the accuracy of the SSD model for small target d...

AI Technical Summary

Problems solved by technology

[0009] The purpose of the present invention is to solve the above problems, provide a small target detection method based on convolutional neural network, strengthen low-level features, reduce redundant extraction, simplify the detection process, improve detection speed, and accurately detect small objects in the image in real time. target, which solves the problem of poor detection effect of small targets in the prior art

Images