【Jiaqin Comment】The blind spot pedestrian detection scheme and technology invented by Hongquan IoT can accurately and real-time detect the dangerous area of the blind area of the driving field of vision, and alarm prompt the pedestrian when the pedestrian enters the blind area danger area, so as to reduce the occurrence of safety accidents.
Set micro network news, as we all know, in the existing vehicle structure design scheme, due to the design of the vehicle itself, there are many blind spots around the body that the mirror cannot reach, and when pedestrians enter the blind area of the driving vision, especially at the time of vehicle turning, there will be great safety hazards.
At present, the monitoring algorithm for blind spots has gradually become popular in vehicles, but the area covered by the camera's viewing angle is still larger than the blind spot. Some existing blind spot detection schemes, because they do not judge some abnormal behaviors, such as green belts, pedestrians outside the barrier do not actually exist in the danger zone, and if the danger area in the viewing angle is not restricted, it will lead to frequent blind spot pedestrian alarms, but will have an impact on the driver.
To this end, on October 11, 2021, Hongquan Internet of Things applied for an invention patent called "Blind Area Pedestrian Detection Method and System" (application number: 202111183040.0), and the applicant was Hangzhou Hongquan Internet of Things Technology Co., Ltd.
According to the relevant information currently published in the patent, let's take a look at this technical solution.
As shown in the figure above, is a schematic diagram of the process of the blind spot pedestrian detection method invented in the patent, in this scheme, the road surface image collected by the target camera is first detected to obtain the position of the pedestrian; secondly, the road surface image is input to the preset semantic segmentation network to obtain the pavement segmentation map, and different blind spots are determined according to the pavement segmentation map; finally, the blind spot where the pedestrian is located is determined according to the position of the pedestrian, and the pedestrian in different blind areas is alerted.
In this scheme, deep learning technology, such as semantic segmentation network model, is mainly used to process the images collected by the camera, such as the UNet network module, which infers the road boundary of road edge segmentation such as road teeth, isolation fences, and green belts, so as to obtain a road surface segmentation map, and divide different blind area danger areas according to the road surface segmentation map.
The application scenario diagram of the blind spot pedestrian detection method of the scheme is shown in the above figure, the target camera sends the road surface image collected during the driving process to the vehicle terminal, and the vehicle terminal receives the road surface image collected by the target camera, and detects the received road surface image, so as to complete the positioning of the pedestrian in the road surface image.
In order to ensure that blind spot monitoring can also work normally in dark scenes, the camera with infrared function is used in this scheme for image acquisition. When the lighting conditions are insufficient, the camera will turn on the infrared fill function to ensure that the road surface image can be collected normally. At the same time, the camera will be installed in the blind spot of the driving field of vision to prevent interference with driving.
As shown above, the structure diagram of the blind spot pedestrian detection system provided in the patent, which mainly includes: pedestrian detection module 310, blind spot determination module 311 and alarm prompt module 312. The pedestrian detection module can detect the road surface image collected by the target camera to obtain the position of the pedestrian; the blind spot determination module can input the road surface image into the preset semantic segmentation network, so as to determine different blind areas according to the segmentation map of the road surface; finally, the alarm prompt module to warn pedestrians in different blind areas.
The above is the blind spot pedestrian detection scheme and technology invented by Hongquan Internet of Things, which can accurately and real-time detect the dangerous area of the blind area of the driving vision, and alarm prompt the pedestrian when the pedestrian enters the blind area danger area, so as to reduce the occurrence of safety accidents.