Object Detection And Feature Recognition Algorithm And Its Application In Construction Safety

The construction site is usually complex and changing,and lead to construction safety accidents.Many of these accidents are caused by workers being hit by mobile devices on the construction site.Therefore,it is necessary to add additional safety measures to protect the construction personnel and to remotely locate the workers and dangerous areas on the construction site.It is necessary to supervise the workers to wear high visibility clothes on the site according to the existing safety specifications and standards to protect the safety of the construction personnel.In the current construction industry,there is no algorithm to detect construction workers and dangerous areas and to identify the characteristics of construction workers at the same time.Therefore,this paper proposes an algorithm of object detection and feature recognition,which uses a unified framework to train a convolutional neural network algorithm with multiple tasks to complete the above tasks.The algorithm reminds construction workers of wearing safety equipment,and protects the safety of construction workers.The main research work of this paper is as follows:First,the object detection and feature recognition algorithm proposed in this paper is a deep learning algorithm based on a large number of image data training.The effect of the algorithm has a great relationship with the distribution of training data and the actual data used.Therefore,an image enhancement algorithm based on dark channel prior and CLAHE is proposed to make all image data meet the outdoor sunny and natural conditions.The training data and actual data have the same distribution,so as to improve the effect of object detection and feature recognition algorithm.Secondly,this paper proposes the structure of feature extraction and utilization based on a convolution neural network.A scale-invariant feature pyramid pooling structure is designed to broaden the features of the input image in the deep network.The structure helps the network better extract multi-scale deep level features in the high-resolution input image.A special feature transmission network is designed to transmit features between networks so that one network can benefit from another network.The feature pyramid structure with constant scale and feature transmission network improve the feature extraction and utilization effect of the deep learning networks.Thirdly,this paper proposes the algorithm of object detection and feature recognition based on a convolution neural network,which can be used to detect an object and recognize features such as construction workers,dangerous areas and characteristics of construction workers.In view of the above tasks,object detection and feature recognition algorithm is added on the basis of YOLOv3,and feature pyramid structure with a constant scale and a feature transmission network are used to improve the network prediction effect.The algorithm of object detection and feature recognition based on a convolutional neural network is an end-to-end multi-task framework,which can meet multiple detection and recognition tasks at the same time.