Small Object Detection Method And Research Based On Deep Learning

In computer vision,small objects are objects that have low resolution and low contrast in images or video.Small objects detection is to locate and identify those small objects in images or video.Because of the small objects in the whole image or video frame occupy space size is small,aggravated the difficulty of this type of objects detection,so the small objects detection is a big challenge in the field of computer vision.The existing small objects detection can improve the accuracy at the cost of computing.In recent years,artificial intelligence has made great progress and has been applied to various fields.Artificial intelligence is also used in the field of computer vision.This paper expounds the development of small object detection based on deep learning,and focuses on the problems existing in small object detection,analyzes the factors affecting small object detection and related knowledge,and analyzes the research on small object detection algorithms at home and abroad.Based on this,we design a small object detection model based on YOLOv3 target detection algorithm.Through the adjustment of the network structure and the optimization of related parameters,the method can identify and locate small object in the image by one process,which effectively improves the recognition rate of small objects in image detection.Based on this,we also design a context detection model of object and object,and the context information between the objects in the image is added.When the target classification is performed after YOLOv3 passes through the convolutional network,the target probability of the relevant category is changed according to the context matrix obtained by the training.This method has played a certain role in the detection of small targets in the experiment,and improved the recognition rate of small objects.