A method for counting the number of people in a bus scene

Abstract

The invention discloses a method for counting the number of people under a bus scene, comprising the following steps of: establishing a passenger head target data set when the passengers getting on and off a bus and marking the data set; training the annotated dataset by a neural network, and detecting the on and off videos of bus by the result of training in order to realize the detection and recognition of passenger targets; on the basis of detecting and recognizing the passenger targets, obtaining the moving trajectory of the passenger target by using a method of matching the distance of the target frame, predicting the position of the passenger target when the passenger target is lost, and carrying out the template matching around the predicted position to obtain the accurate positioninformation; and eliminating the false target trajectory according to the confidence level and direction information of the passenger target in the target trajectory so as to realize the statistics ofthe number of passengers getting on and off the bus. The method of the invention has the characteristics of low cost, small influence by external factors, relatively small calculation amount, good robustness and accurate identification effect under complex and changeable scenes.

Description

technical field [0001] The invention relates to the field of image processing and counting, in particular to a method for counting people in a bus scene. Background technique [0002] Bus passenger flow is an important data source for public transport management departments to optimize urban bus routes and vehicle scheduling. How to obtain accurate bus passenger flow data has important research value. [0003] People counting technologies such as pressure sensors, infrared sensors, and thermal imagers have appeared in bus passenger flow counting. However, there are many defects and problems in these passenger flow counting methods, and they cannot effectively adapt to complex and crowded bus scenes. For example, the pressure sensor counts the number of people by sensing the pressure and changes of the pedals, which has the problems of easy damage, inconvenient daily maintenance, and high cost; infrared products cannot solve the problem of occlusion when passengers are crowd...

AI Technical Summary

Problems solved by technology

This type of camera can obtain the depth information of the scene, and pedestrians have relatively stable characteristics in three-dimensional space, so spatial information can be used to identify pedestrians, but it also has disadvantages such as high equipment cost, being easily affected by strong light, and a large amount of calculation.

In addition, using traditional image processing methods for pedestrian detection and recognition requires manual extraction of features and training of classifiers.

However, artificially extracted features are often less robust and difficult to adapt to complex and changeable scenes.

Images