Bus passenger flow statistics method and system based on deep learning

Abstract

The invention discloses a bus passenger flow statistics method based on deep learning, which comprises the following steps: offline collecting passenger flow sample data, and carrying out format conversion; building a deep learning model for training; testing the trained model by using a test sample, and storing learned model parameters; detecting the real-time video stream by adopting the learnedmodel, and storing the detected head information; tracking the continuous multi-frame detected head information, and determining a plurality of moving targets; and comparing the determined moving target with a preset counting threshold, and PERFORMING counting when the moving target reaches the counting threshold. During head detection, secondary verification is carried out on a detected target frame, the target frame is input into a pre-trained classification network, classification confirmation is carried out on the target frame to determine whether the target frame is the head or not, thedetection rate of the head is greatly improved, the missing detection rate and the false detection rate of the head are reduced, and the problems of congestion, knapsacks, caps and the like are well solved.

Description

technical field [0001] The invention relates to a bus passenger flow statistics system, in particular to a bus passenger flow statistics method and system based on deep learning. Background technique [0002] At present, there are mainly three kinds of common passenger flow statistics techniques: [0003] (1) Manual statistical method [0004] Manual statistics mainly record the number of passengers getting on and off at each station manually by the on-board staff, and perform manual summary statistics at the terminal. The real-time number of visitors on and off the site. [0005] (2) IC card-based method [0006] IC cards store a large amount of personal information of passengers. This method records the ID of each passenger who swipes the card for passenger flow statistics. The card is not connected to the Internet, so it is impossible to realize online real-time statistics of vehicle passenger flow. [0007] (3) Automatic counting method of passenger flow based on im...

AI Technical Summary

Problems solved by technology

Among them, the image vision processing technology mainly detects and tracks the heads of people getting on and off the bus, and realizes the automatic counting of passenger flow. However, the accuracy of the traditional machine learning image processing method is greatly affected by the light, and it cannot solve the problems of crowding, hats, backpacks, etc.

Images