CN112581748A - Garage traffic-out pedestrian safety prompting system based on convolutional neural network - Google Patents
Garage traffic-out pedestrian safety prompting system based on convolutional neural network Download PDFInfo
- Publication number
- CN112581748A CN112581748A CN202011512932.6A CN202011512932A CN112581748A CN 112581748 A CN112581748 A CN 112581748A CN 202011512932 A CN202011512932 A CN 202011512932A CN 112581748 A CN112581748 A CN 112581748A
- Authority
- CN
- China
- Prior art keywords
- pedestrian
- vehicle
- garage
- unit
- detection unit
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000013527 convolutional neural network Methods 0.000 title claims description 39
- 238000001514 detection method Methods 0.000 claims abstract description 104
- 238000012545 processing Methods 0.000 claims description 43
- 238000012549 training Methods 0.000 claims description 36
- 238000004891 communication Methods 0.000 claims description 24
- 238000005286 illumination Methods 0.000 claims description 12
- 238000000034 method Methods 0.000 claims description 5
- 238000013459 approach Methods 0.000 claims description 3
- 230000036544 posture Effects 0.000 claims description 3
- SAZUGELZHZOXHB-UHFFFAOYSA-N acecarbromal Chemical compound CCC(Br)(CC)C(=O)NC(=O)NC(C)=O SAZUGELZHZOXHB-UHFFFAOYSA-N 0.000 claims 2
- 238000013528 artificial neural network Methods 0.000 claims 2
- 238000007781 pre-processing Methods 0.000 abstract description 3
- 238000000605 extraction Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 230000026676 system process Effects 0.000 description 2
- 239000003086 colorant Substances 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000001960 triggered effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/005—Traffic control systems for road vehicles including pedestrian guidance indicator
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/017—Detecting movement of traffic to be counted or controlled identifying vehicles
- G08G1/0175—Detecting movement of traffic to be counted or controlled identifying vehicles by photographing vehicles, e.g. when violating traffic rules
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/02—Detecting movement of traffic to be counted or controlled using treadles built into the road
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/09—Arrangements for giving variable traffic instructions
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/09—Arrangements for giving variable traffic instructions
- G08G1/095—Traffic lights
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Traffic Control Systems (AREA)
- Image Analysis (AREA)
Abstract
The pressure detection unit is used for detecting whether vehicles are about to exit from the garage or not and triggering the three camera devices to be turned on, wherein the first camera device and the second camera device are respectively used for shooting the conditions of pedestrians on the left side and the right side of a road adjacent to the garage exit, and the third camera device is used for shooting the conditions of the vehicles on the road inside the garage exit. Firstly, deformation preprocessing is carried out on the image, then, the moving direction and the distance relative to the garage exit of the pedestrian are obtained through a pedestrian detection unit, and the moving direction and the license plate number of the vehicle are obtained through a vehicle detection unit. When the pedestrian detecting unit detects that there is the pedestrian in garage export both sides, this pedestrian moves towards garage exit position simultaneously and when this pedestrian is located within the predetermined safe distance scope, the vehicle that the system is about to roll off from the garage through the pilot lamp and the audio amplifier of garage exit position reminds, simultaneously to the on-vehicle wiFi equipment propelling movement message of vehicle to this in time tells the car owner that there is the pedestrian to be about to be close to the garage export, makes the car owner in time dodge.
Description
Technical Field
The invention relates to the technical field of intelligent garages, in particular to a garage vehicle-out pedestrian safety prompting system based on a convolutional neural network.
Background
The exit of the underground garage is often a sharp turn and a steep slope, and the speed of the vehicle is not well controlled.
The exit direction of the partial garage is vertical to and closely connected with the direction of the adjacent road. Pedestrians or vehicles can not notice vehicles coming out of the garage when passing through the garage exit, and certain potential safety hazards exist for the pedestrians and the vehicles passing in front of the garage.
When a driver drives out of the garage, vision blind areas exist on the left side and the right side of the exit position of the garage for the driver, and the driver can hardly observe pedestrians and vehicles on the two sides of the exit of the garage. In the prior art, when a driver drives a vehicle out of a garage, an intelligent garage management system cannot inform a vehicle owner whether a pedestrian is about to approach an exit of the garage or not in time, and the vehicle owner cannot decelerate and avoid in time.
Disclosure of Invention
In order to overcome the defects in the prior art, the invention detects whether a vehicle is about to exit from the garage through the pressure detection unit and triggers the opening of the three camera devices, wherein the first camera device and the second camera device are respectively used for shooting the conditions of pedestrians on the left side and the right side of a road adjacent to the exit of the garage, and the third camera device is used for shooting the conditions of vehicles on the road inside the exit of the garage. Firstly, deformation preprocessing is carried out on the image, then, the moving direction and the distance relative to the garage exit of the pedestrian are obtained through a pedestrian detection unit, and the moving direction and the license plate number of the vehicle are obtained through a vehicle detection unit. When pedestrian detecting element detects that there is the pedestrian in garage export both sides, this pedestrian moves towards garage exit position simultaneously and when this pedestrian is located within the predetermined safe distance scope, the vehicle that the system was about to roll off from the garage through the pilot lamp and the audio amplifier of garage exit position reminds, simultaneously to the on-vehicle wiFi equipment propelling movement message of vehicle to this in time tells the car owner that there is the pedestrian to be about to be close to the garage export, makes the car owner in time slow down and dodge, with this security that improves the driving.
The invention is realized by adopting the following technical scheme, and the garage vehicle-out pedestrian safety prompting system based on the convolutional neural network designed according to the purpose comprises: the device comprises a pressure detection unit, an image acquisition unit, a server, an image processing unit, a pedestrian safety prompt triggering unit, a driving unit, a prompting unit, a storage unit and a wireless communication unit.
The prompting unit comprises: the illumination prompting unit and the voice prompting unit.
The pressure detection unit and the image acquisition unit are used as input units and are connected with the server; the server is connected with the image processing unit, the pedestrian safety prompt triggering unit, the wireless communication unit and the storage unit; the output end of the server is connected with the input end of the driving unit; the output end of the driving unit is connected with the illumination prompting unit and the voice prompting unit.
The pressure detection unit is composed of a plurality of pressure sensors and is placed at a road position inside the garage, which is 10-20 meters away from the garage exit. The pressure detection unit is used for detecting and judging whether a vehicle is about to run out of the garage or not, and if the pressure detection unit detects that the vehicle is about to run out of the garage, a camera shooting starting trigger signal is generated. If the pressure detection unit detects that the current pressure value changes and the current pressure value is larger than the preset pressure value, the fact that the vehicle is about to be driven out of the garage is judged, and a camera shooting starting trigger signal is generated.
A convolutional neural network is adopted, and a pedestrian model, a license plate model and various vehicle models are obtained by learning and training a large number of pedestrian pictures, license plate pictures and vehicle pictures.
During training, 50000 pedestrian pictures with different postures are adopted to train a pedestrian model, 50000 license plate pictures are adopted to train the license plate model, and 50000 vehicle pictures with different types are respectively adopted to train various vehicle models.
An image acquisition unit: and three camera devices, namely a first camera device, a second camera device and a third camera device, are fixedly installed at the top end of the garage outlet. The first camera device is used for shooting the pedestrian condition on the left side of the road adjacent to the garage outlet, the second camera device is used for shooting the pedestrian condition on the right side of the road adjacent to the garage outlet, and the third camera device is used for shooting the vehicle condition on the road inside the garage outlet. The three camera devices acquire a preset number of continuous images in real time; the camera shooting device is one or a combination of a non-infrared camera shooting device and an infrared camera shooting device.
The server is used for data processing.
The image processing unit includes: the image deformation processing unit, the pedestrian detection unit and the vehicle detection unit. Wherein, pedestrian detection unit includes: a pedestrian direction detection unit and a pedestrian distance detection unit; the vehicle detection unit includes: a vehicle direction detecting unit and a license plate number detecting unit.
An image deformation processing unit: and carrying out deformation processing on the continuous images of the preset number acquired by the image acquisition unit in real time.
A pedestrian detection unit: processing the continuous images processed by the image deformation processing unit and judging whether pedestrians exist in the images; the image is subjected to feature extraction and is matched with pedestrian model features obtained through convolutional neural network training, if the pedestrian features in the image are matched with the pedestrian model features obtained through convolutional neural network training, it is judged that pedestrians exist in the image, and then the pedestrians in the image are marked.
Pedestrian direction detection means: and calculating the continuous images processed by the pedestrian detection unit and determining the moving direction of the pedestrian relative to the garage outlet so as to determine whether the pedestrian is about to approach the exit position of the garage. If the moving speed of the pedestrian relative to the garage exit is positive, the pedestrian is close to the garage exit.
Pedestrian distance detection unit: and calculating the continuous images processed by the pedestrian detection unit so as to determine the distance of the pedestrian relative to the garage exit.
A vehicle detection unit: processing the continuous images processed by the image deformation processing unit and judging whether vehicles exist in the images or not; and if the features of the vehicle in the image are matched with the features of the vehicle model obtained by the convolutional neural network training, judging that the vehicle exists in the image and marking the vehicle in the image.
Vehicle direction detection unit: and calculating the continuous images processed by the vehicle detection unit, determining the driving direction of the vehicle relative to the garage exit, and further determining whether the vehicle is about to exit from the garage. In any two frames of images, the difference value of the distance of the vehicle marked by the vehicle detection unit relative to the garage exit is divided by the time for acquiring the two frames of images to obtain the moving speed of the vehicle relative to the garage exit, and if the moving speed of the vehicle relative to the garage exit is positive, the vehicle is about to run out of the garage.
License plate number detection unit: the vehicle marked by the vehicle detection unit is processed to identify the license plate number of the marked vehicle.
Pedestrian safety suggestion trigger unit: when the pedestrian detection unit detects that people exist at two sides of the garage outlet, the pedestrian direction detection unit detects that the pedestrian moves towards the garage outlet, and the pedestrian distance detection unit detects that the pedestrian is located within the preset safe distance range, the pedestrian safety prompt triggering unit executes pedestrian safety prompt operation and pedestrian safety message pushing operation.
And (3) pedestrian safety prompting operation: the server adjusts one or more modes of brightness, flicker frequency, color, on-off state or voice broadcast of the first prompt lamp, the second prompt lamp and the third prompt lamp close to the garage exit position through the driving unit to perform safety prompt and guide on the car owner at the garage exit position.
Pedestrian safety message pushing operation: the server firstly obtains the license plate number of the current vehicle through the license plate number detection unit, and then reads the license plate number-vehicle-mounted WiFi _ ID corresponding table information stored in the storage unit according to the license plate number of the current vehicle so as to determine the vehicle-mounted WiFi _ ID information corresponding to the license plate number of the current vehicle. And then, the server carries out wireless communication with the vehicle-mounted WiFI equipment of the current vehicle through the wireless communication unit and sends pedestrian data information on two sides of the garage outlet. The vehicle-mounted WiFI equipment of the current vehicle analyzes the data information after receiving the data information sent by the wireless communication unit, and prompts a vehicle owner through voice. Therefore, when pedestrians are close to the garage outlet on the two sides of the garage outlet, safety prompt is timely carried out on the vehicle owner about to exit the garage.
The driving unit is used for driving the illumination prompting unit and the voice prompting unit.
The illumination prompt unit includes: the first warning light, the second warning light, the third warning light and the light guide plate. The prompting lamp is a multicolor LED lamp; each prompting lamp is fixed in one light guide plate respectively; the light guide plate is rectangular, and patterns are printed on the bottom surface of the light guide plate; the light guide plate reflects light and colors of the indicator light uniformly. The first prompting lamp is fixed on a road in the garage 5-10 meters away from the garage outlet, the second prompting lamp is fixed on a wall on the left side in the garage 1-5 meters away from the garage outlet, and the third prompting lamp is fixed on a wall on the right side in the garage 1-5 meters away from the garage outlet. The illumination prompting unit carries out light prompting and guiding on the vehicle owner at the garage exit position by adjusting the brightness, the flashing frequency, the color and the on-off state of the prompting lamp.
The voice prompt unit includes: the voice generating unit, the sound box and the corpus are connected; the loudspeaker box is fixed right above the second prompt lamp and the third prompt lamp. The voice prompt unit carries out voice prompt and guide on the car owner at the garage exit position through voice broadcast.
The storage unit is used for accessing and storing license plate number-vehicle-mounted WiFi _ ID corresponding table information and pedestrian models, license plate models and various vehicle model information obtained through convolutional neural network training.
The wireless communication unit is a WiFi module and is used for carrying out wireless communication with the vehicle-mounted WiFi receiving equipment.
The program flow of the present invention is as follows.
Step S11: a convolutional neural network is adopted, and a pedestrian model, a license plate model and various vehicle models are obtained by learning and training a large number of pedestrian pictures, license plate pictures and vehicle pictures.
Step S12: and the pressure detection unit 10-20 meters away from the garage outlet detects the current pressure value in real time.
Step S13: judging whether a vehicle is about to run out of the garage or not; if the pressure detection unit detects that the current pressure value changes and the current pressure value is larger than the preset pressure value, it is determined that a vehicle is about to exit from the garage, and at this time, step S14 is executed; otherwise, the process returns to step S12.
Step S14: three camera devices fixedly mounted at the top end of the garage outlet are opened, and the three camera devices acquire continuous images of preset quantity in real time. The first camera device is used for shooting the pedestrian condition on the left side of the road adjacent to the garage outlet, the second camera device is used for shooting the pedestrian condition on the right side of the road adjacent to the garage outlet, and the third camera device is used for shooting the vehicle condition on the road inside the garage outlet.
Step S15: and carrying out deformation processing on the preset number of continuous images acquired by the three camera devices in real time through the image deformation processing unit.
Step S16: and processing the image processed by the image deformation processing unit through a pedestrian detection unit and judging whether a pedestrian exists in the image. Extracting the features of the image and matching the features with the pedestrian model features obtained by the convolutional neural network training, if the features of the pedestrian in the image are matched with the features of the pedestrian model obtained by the convolutional neural network training, judging that the pedestrian exists in the image, and then executing the step S17; otherwise, the execution returns to step S12.
Step S17: firstly, the pedestrian in the image is marked, then, the moving direction of the marked pedestrian in the image is calculated and judged through the pedestrian direction detection unit, and the distance of the marked pedestrian in the image relative to the garage exit is calculated through the pedestrian distance detection unit.
Step S18: and processing the image processed by the image deformation processing unit through a vehicle detection unit and judging whether a vehicle exists in the image. Extracting the features of the image, matching the extracted features with various vehicle model features obtained by training a convolutional neural network, judging that a vehicle exists in the image if the features of the vehicle in the image are matched with the vehicle model features obtained by training the convolutional neural network, and then executing step S19; otherwise, the execution returns to step S12.
Step S19: firstly, the vehicle in the image is marked, then the moving direction of the marked vehicle in the image is calculated and judged through the vehicle direction detection unit, and the license plate number of the marked vehicle is identified and judged through the license plate number detection unit.
Step S20: when the pedestrian detection unit detects that there is the pedestrian in garage export both sides, pedestrian direction detection unit detects this pedestrian and moves towards garage exit position and pedestrian apart from the detection unit when detecting that this pedestrian is located predetermined safe distance within range simultaneously, pedestrian safety suggestion trigger unit carries out pedestrian safety suggestion operation and pedestrian safety message propelling movement operation.
Step S21: and (3) pedestrian safety prompting operation: the server adjusts one or more modes of brightness, flicker frequency, color, on-off state or voice broadcast of the first prompt lamp, the second prompt lamp and the third prompt lamp close to the garage exit position through the driving unit to perform safety prompt and guide on the car owner at the garage exit position.
Step S22: pedestrian safety message pushing operation: the server firstly obtains the license plate number of the current vehicle through the license plate number detection unit, and then reads the license plate number-vehicle-mounted WiFi _ ID corresponding table information stored in the storage unit so as to determine the vehicle-mounted WiFi _ ID information corresponding to the license plate number of the current vehicle. And then, the server sends the pedestrian data information on two sides of the garage outlet to the vehicle-mounted WiFI equipment of the current vehicle through the wireless communication unit.
Step S23: the vehicle-mounted WiFI equipment of the current vehicle analyzes the data information after receiving the data information sent by the wireless communication unit, and carries out pedestrian safety prompt on the vehicle owner through voice.
Drawings
FIG. 1 is a block diagram of the system of the present invention.
Fig. 2 is a schematic diagram of a scene application of the present invention.
Fig. 3 is a schematic view of the shooting angle ranges of the first and second imaging devices in the present invention.
FIG. 4 is a flowchart of the process of the present invention.
In fig. 2, 1 is a garage, 2 is a garage exit passage, 3 is a garage exit adjacent street, 11 is a pressure detection unit, 12 is a first camera device, 13 is a second camera device, 14 is a third camera device, 15 is a first warning light, 16 is a second warning light, and 17 is a third warning light.
In fig. 3, w1 is the shooting angle range of the first image pickup device, and w2 is the shooting angle range of the second image pickup device.
Detailed Description
The invention is further illustrated with reference to the accompanying drawings and specific examples.
A first embodiment, as shown in fig. 1-4.
The invention detects and judges whether a vehicle is about to run out of the garage through the 11 pressure detection units, and if the 11 pressure detection units detect that the vehicle is about to run out of the garage according to the pressure change, the 11 pressure detection units generate a camera shooting starting trigger signal.
One functional implementation of the camera start trigger signal generated by the present invention is as follows.
The camera shooting starting trigger signal is used for triggering three camera shooting devices for opening the garage exit position. The system comprises a garage exit, a first camera device 12, a second camera device 13, a third camera device 14 and a third camera device 14, wherein the first camera device and the second camera device 13 are respectively used for shooting conditions of pedestrians on the left side and the right side of a road adjacent to the garage exit, and the third camera device 14 is used for shooting conditions of vehicles on the road inside the garage exit. When no vehicle exits from the garage, the three camera devices at the exit position of the garage are kept in a closed state, so that energy consumption is saved. When a vehicle is about to run out of the garage, the three camera devices at the exit position of the garage are triggered to be opened, the three camera devices acquire video image information of the exit position of the garage in real time, the system processes the image information acquired by the three camera devices in real time, identifies information such as a pedestrian state, a vehicle state and a license plate number in an image, and judges and executes subsequent pedestrian safety prompting operation and pedestrian safety message pushing operation.
Another functional implementation of the image capture start trigger signal generated by the present invention is as follows.
The three camera devices installed at the exit position of the garage can be always kept in an open state and are used as one part of the intelligent garage monitoring and management system, and the three camera devices collect video image information of the exit position of the garage in real time. If the pressure detection unit 11 generates a camera shooting starting trigger signal, the system processes the acquired image information, identifies information such as pedestrian information, vehicle information and license plate numbers in the image, and judges and executes subsequent pedestrian safety prompting operation and pedestrian safety message pushing operation. If the pressure detection unit 11 does not generate a camera shooting starting trigger signal, the system does not process the acquired image information and does not execute related subsequent operations.
A second embodiment, as shown in fig. 1-4.
A convolutional neural network is adopted, and a pedestrian model, a license plate model and various vehicle models are obtained by learning and training a large number of pedestrian pictures, license plate pictures and vehicle pictures. During training, 50000 pedestrian pictures with different postures are adopted to train a pedestrian model, 50000 license plate pictures are adopted to train the license plate model, and 50000 vehicle pictures with different types are respectively adopted to train various vehicle models. Therefore, each model is trained through a large number of training sets, and the recognition accuracy is increased.
As shown in fig. 3, the image acquisition unit: the three camera devices are fixedly installed at the top end of the garage outlet, and a preset number of continuous images are obtained in real time, wherein w1 is the shooting angle range of the first camera device, and w2 is the shooting angle range of the second camera device.
The system firstly carries out deformation processing on the image through the image processing unit so that the image meets the preprocessing requirement.
Then, the images shot by the 12 first camera device and the 13 second camera device are subjected to feature extraction through the convolutional neural network and are matched with a pedestrian model obtained through training of the convolutional neural network, and whether pedestrians exist in the images shot by the 12 first camera device and the 13 second camera device or not is judged. If the image contains the pedestrian, firstly, marking the pedestrian in the image, and then calculating and judging the moving direction and the distance of the marked pedestrian relative to the garage exit.
And then, performing feature extraction on the image shot by the 14 third camera device through the convolutional neural network, matching the image with various vehicle models obtained by training of the convolutional neural network, and judging whether a vehicle exists in the image. If the image contains the vehicle, firstly, the vehicle in the image is marked, and then the running direction and the license plate number of the marked vehicle are calculated and judged.
When pedestrian detecting element detects garage export both sides and has the pedestrian, this pedestrian moves towards garage exit position simultaneously and when this pedestrian is located predetermined safe distance within the scope, the server passes through drive unit drive illumination suggestion unit and voice prompt unit and carries out light sum voice prompt to the vehicle of garage exit position, the simultaneous system passes through wireless communication unit to the on-vehicle wiFi equipment propelling movement message of vehicle to this in time informs the car owner that there is the pedestrian to be close to the garage export, make the car owner in time slow down and dodge, with this security that improves the driving.
In any two frames of images, the moving speed of the pedestrian relative to the garage exit = the difference of the distances of the same pedestrian marked by the pedestrian detection unit relative to the garage exit/the time for acquiring the two frames of images. If the moving speed of the pedestrian relative to the garage exit is positive, the pedestrian is close to the garage exit.
In any two frames of images, the moving speed of the vehicle relative to the garage exit = the difference of the distances of the same vehicle marked by the vehicle detection unit relative to the garage exit/the time for acquiring the two frames of images. If the moving speed of the vehicle relative to the garage exit is positive, the vehicle is about to exit from the garage.
A third embodiment, as shown in fig. 1-3.
The illumination prompting unit and the voice prompting unit perform light and voice prompting on the vehicle at the garage exit position in five conditions as follows.
If no vehicle is driven out of the garage, the system does not remind the vehicle.
When a vehicle is about to run out of the garage and no pedestrian exists on the left side and the right side of the road adjacent to the garage exit, the system does not remind the vehicle.
When a vehicle is about to roll out of the garage, when a person is on the left side of the adjacent road of the garage exit, the first prompt lamp 15 on the left side of the garage exit position and the third prompt lamp 17 on the inside road of the garage are lightened and quickly flicker to emit red light, the sound box in the voice prompt unit carries out voice broadcast, and meanwhile, the system pushes messages to the vehicle-mounted WiFi equipment of the vehicle through the wireless communication unit.
When a vehicle is about to roll out of the garage, when a person is on the right side of the adjacent road of the garage exit, the 16 second prompt lamps on the right side of the garage exit position and the 17 third prompt lamps on the road inside the garage are lightened and quickly flicker to emit red light, the sound boxes in the voice prompt units are subjected to voice broadcasting, and meanwhile, the system pushes messages to the vehicle-mounted WiFi equipment of the vehicle through the wireless communication unit.
When a vehicle is about to roll out of the garage, when people exist on the left side and the right side of a road adjacent to the garage exit, a first prompt lamp 15 on the left side of the garage exit position, a second prompt lamp 16 on the right side of the garage exit position and a third prompt lamp 17 on the road inside the garage are lightened and flash fast to emit red light, a sound box in a voice prompt unit conducts voice broadcast, and meanwhile, the system pushes messages to vehicle-mounted WiFi equipment of the vehicle through a wireless communication unit.
A fourth embodiment, as shown in fig. 1, 2 and 4.
The program flow of the present invention is as follows.
Step S11: a convolutional neural network is adopted, and a pedestrian model, a license plate model and various vehicle models are obtained by learning and training a large number of pedestrian pictures, license plate pictures and vehicle pictures.
Step S12: and the pressure detection unit 10-20 meters away from the garage outlet detects the current pressure value in real time.
Step S13: judging whether a vehicle is about to run out of the garage or not; if the pressure detection unit detects that the current pressure value changes and the current pressure value is larger than the preset pressure value, it is determined that a vehicle is about to exit from the garage, and at this time, step S14 is executed; otherwise, the process returns to step S12.
Step S14: three camera devices fixedly mounted at the top end of the garage outlet are opened, and the three camera devices acquire continuous images of preset quantity in real time. The first camera device is used for shooting the pedestrian condition on the left side of the road adjacent to the garage outlet, the second camera device is used for shooting the pedestrian condition on the right side of the road adjacent to the garage outlet, and the third camera device is used for shooting the vehicle condition on the road inside the garage outlet.
Step S15: and carrying out deformation processing on the preset number of continuous images acquired by the three camera devices in real time through the image deformation processing unit.
Step S16: and processing the image processed by the image deformation processing unit through a pedestrian detection unit and judging whether a pedestrian exists in the image. Extracting the features of the image and matching the features with the pedestrian model features obtained by the convolutional neural network training, if the features of the pedestrian in the image are matched with the features of the pedestrian model obtained by the convolutional neural network training, judging that the pedestrian exists in the image, and then executing the step S17; otherwise, the execution returns to step S12.
Step S17: firstly, the pedestrian in the image is marked, then, the moving direction of the marked pedestrian in the image is calculated and judged through the pedestrian direction detection unit, and the distance of the marked pedestrian in the image relative to the garage exit is calculated through the pedestrian distance detection unit.
Step S18: and processing the image processed by the image deformation processing unit through a vehicle detection unit and judging whether a vehicle exists in the image. Extracting the features of the image, matching the extracted features with various vehicle model features obtained by training a convolutional neural network, judging that a vehicle exists in the image if the features of the vehicle in the image are matched with the vehicle model features obtained by training the convolutional neural network, and then executing step S19; otherwise, the execution returns to step S12.
Step S19: firstly, the vehicle in the image is marked, then the moving direction of the marked vehicle in the image is calculated and judged through the vehicle direction detection unit, and the license plate number of the marked vehicle is identified and judged through the license plate number detection unit.
Step S20: when the pedestrian detection unit detects that there is the pedestrian in garage export both sides, pedestrian direction detection unit detects this pedestrian and moves towards garage exit position and pedestrian apart from the detection unit when detecting that this pedestrian is located predetermined safe distance within range simultaneously, pedestrian safety suggestion trigger unit carries out pedestrian safety suggestion operation and pedestrian safety message propelling movement operation.
Step S21: and (3) pedestrian safety prompting operation: the server adjusts one or more modes of brightness, flicker frequency, color, on-off state or voice broadcast of the first prompt lamp, the second prompt lamp and the third prompt lamp close to the garage exit position through the driving unit to perform safety prompt and guide on the car owner at the garage exit position.
Step S22: pedestrian safety message pushing operation: the server firstly obtains the license plate number of the current vehicle through the license plate number detection unit, and then reads the license plate number-vehicle-mounted WiFi _ ID corresponding table information stored in the storage unit so as to determine the vehicle-mounted WiFi _ ID information corresponding to the license plate number of the current vehicle. And then, the server sends the pedestrian data information on two sides of the garage outlet to the vehicle-mounted WiFI equipment of the current vehicle through the wireless communication unit.
Step S23: the vehicle-mounted WiFI equipment of the current vehicle analyzes the data information after receiving the data information sent by the wireless communication unit, and carries out pedestrian safety prompt on the vehicle owner through voice.
The above description is only for the preferred embodiment of the present invention, and is not intended to limit the scope of the present invention; all equivalent changes and modifications made according to the present invention are covered by the scope of the claims of the present invention.
Claims (4)
1. The utility model provides a garage pedestrian safety prompting system that goes out a car based on convolution neural network which characterized in that, garage pedestrian safety prompting system that goes out a car based on convolution neural network includes: the device comprises a pressure detection unit, an image acquisition unit, a server, an image processing unit, a pedestrian safety prompt triggering unit, a driving unit, a prompting unit, a storage unit and a wireless communication unit; the prompting unit comprises: the illumination prompting unit and the voice prompting unit;
the pressure detection unit and the image acquisition unit are used as input units and are connected with the server; the server is connected with the image processing unit, the pedestrian safety prompt triggering unit, the wireless communication unit and the storage unit; the output end of the server is connected with the input end of the driving unit; the output end of the driving unit is connected with the illumination prompting unit and the voice prompting unit;
the pressure detection unit consists of a plurality of pressure sensors and is arranged at a road position inside the garage, which is 10-20 meters away from the garage exit; the pressure detection unit is used for detecting and judging whether a vehicle is about to run out of the garage, and if the pressure detection unit detects that the current pressure value changes and the current pressure value is greater than a preset pressure value, the pressure detection unit judges that the vehicle is about to run out of the garage and generates a camera shooting starting trigger signal;
a convolutional neural network is adopted, and a pedestrian model, a license plate model and various vehicle models are obtained by learning and training a large number of pedestrian pictures, license plate pictures and vehicle pictures;
an image acquisition unit: the method comprises the following steps that three camera devices, namely a first camera device, a second camera device and a third camera device, are fixedly installed at the top end of a garage outlet; the system comprises a first camera device, a second camera device, a third camera device and a third camera device, wherein the first camera device is used for shooting the pedestrian condition on the left side of a road adjacent to a garage exit, the second camera device is used for shooting the pedestrian condition on the right side of the road adjacent to the garage exit, and the third camera device is used for shooting the vehicle condition on the road inside the garage exit; the three camera devices acquire a preset number of continuous images in real time; the camera shooting device is one or a combination of a non-infrared camera shooting device and an infrared camera shooting device;
the server is used for data processing;
the image processing unit includes: an image deformation processing unit, a pedestrian detection unit and a vehicle detection unit; wherein, pedestrian detection unit includes: a pedestrian direction detection unit and a pedestrian distance detection unit; the vehicle detection unit includes: a vehicle direction detecting unit and a license plate number detecting unit;
an image deformation processing unit: carrying out deformation processing on a preset number of continuous images acquired by an image acquisition unit in real time;
a pedestrian detection unit: processing the continuous images processed by the image deformation processing unit and judging whether pedestrians exist in the images; extracting the features of the image and matching the features with the pedestrian model features obtained by convolutional neural network training, if the features of the pedestrian in the image are matched with the features of the pedestrian model obtained by convolutional neural network training, judging that the pedestrian exists in the image, and then marking the pedestrian in the image;
pedestrian direction detection means: calculating continuous images processed by the pedestrian detection unit and determining the moving direction of the pedestrian relative to the garage exit so as to determine whether the pedestrian is about to approach the exit position of the garage; if the moving speed of the pedestrian relative to the garage outlet is positive, the pedestrian is close to the garage outlet;
pedestrian distance detection unit: calculating continuous images processed by the pedestrian detection unit so as to determine the distance of the pedestrian relative to the garage exit;
a vehicle detection unit: processing the continuous images processed by the image deformation processing unit and judging whether vehicles exist in the images or not; extracting the features of the image and matching with various vehicle model features obtained by convolutional neural network training, if the features of the vehicle in the image are matched with the vehicle model features obtained by convolutional neural network training, judging that the vehicle exists in the image and marking the vehicle in the image;
vehicle direction detection unit: calculating continuous images processed by the vehicle detection unit and determining the driving direction of the vehicle relative to the garage exit so as to determine whether the vehicle is about to exit from the garage; in any two frames of images, the difference value of the distance between the vehicle marked by the vehicle detection unit and the garage exit is divided by the time for acquiring the two frames of images to obtain the moving speed of the vehicle relative to the garage exit, and if the moving speed of the vehicle relative to the garage exit is positive, the vehicle is about to run out of the garage;
license plate number detection unit: processing the vehicle marked by the vehicle detection unit, and identifying the license plate number of the marked vehicle;
pedestrian safety suggestion trigger unit: when the pedestrian detection unit detects that pedestrians are arranged on two sides of the garage outlet, the pedestrian direction detection unit detects that the pedestrians move towards the garage outlet, and the pedestrian distance detection unit detects that the pedestrians are located within a preset safety distance range, the pedestrian safety prompt triggering unit executes pedestrian safety prompt operation and pedestrian safety message pushing operation;
and (3) pedestrian safety prompting operation: the server adjusts one or more modes of brightness, flicker frequency, color, on-off state or voice broadcast of a first prompt lamp, a second prompt lamp and a third prompt lamp close to the garage exit position through the driving unit to perform safety prompt and guidance on the car owner at the garage exit position;
pedestrian safety message pushing operation: the server firstly obtains the license plate number of the current vehicle through the license plate number detection unit, and then reads the license plate number-vehicle-mounted WiFi _ ID corresponding table information stored in the storage unit according to the license plate number of the current vehicle so as to determine the vehicle-mounted WiFi _ ID information corresponding to the license plate number of the current vehicle; then, the server carries out wireless communication with vehicle-mounted WiFI equipment of the current vehicle through a wireless communication unit and sends pedestrian data information on two sides of the garage outlet; after receiving the data information sent by the wireless communication unit, the vehicle-mounted WiFI equipment of the current vehicle analyzes the data information and prompts a vehicle owner through voice;
the driving unit is used for driving the illumination prompting unit and the voice prompting unit;
the illumination prompt unit includes: the first prompt lamp, the second prompt lamp, the third prompt lamp and the light guide plate; each prompting lamp is fixed in one light guide plate respectively; the prompting lamp is a multicolor LED lamp; the first prompting lamp is fixed on a road in the garage 5-10 meters away from the garage exit, the second prompting lamp is fixed on a wall on the left side in the garage 1-5 meters away from the garage exit, and the third prompting lamp is fixed on a wall on the right side in the garage 1-5 meters away from the garage exit; the illumination prompting unit carries out light prompting and guiding on the vehicle owner at the garage exit position by adjusting the brightness, the flashing frequency, the color and the on-off state of the prompting lamp;
the voice prompt unit includes: the voice generating unit, the sound box and the corpus are connected; the sound box is fixed right above the second prompt lamp and the third prompt lamp; the voice prompt unit carries out voice prompt and guidance on the car owner at the exit position of the garage through voice broadcast;
the storage unit is used for accessing and storing license plate number-vehicle-mounted WiFi _ ID corresponding table information and pedestrian models, license plate models and various vehicle model information obtained through convolutional neural network training;
the wireless communication unit is a WiFi module and is used for carrying out wireless communication with the vehicle-mounted WiFi receiving equipment.
2. The garage exit pedestrian safety prompting system based on the convolutional neural network as claimed in claim 1, characterized in that the program flow of the invention is as follows:
step S11: a convolutional neural network is adopted, and a pedestrian model, a license plate model and various vehicle models are obtained by learning and training a large number of pedestrian pictures, license plate pictures and vehicle pictures;
step S12: detecting the current pressure value in real time by a pressure detection unit which is 10-20 meters away from the garage outlet;
step S13: judging whether a vehicle is about to run out of the garage or not; if the pressure detection unit detects that the current pressure value changes and the current pressure value is larger than the preset pressure value, it is determined that a vehicle is about to exit from the garage, and at this time, step S14 is executed; otherwise, returning to execute the step S12;
step S14: opening three camera devices fixedly installed at the top end of the garage outlet, and acquiring a preset number of continuous images in real time by the three camera devices; the system comprises a first camera device, a second camera device, a third camera device and a third camera device, wherein the first camera device is used for shooting the pedestrian condition on the left side of a road adjacent to a garage exit, the second camera device is used for shooting the pedestrian condition on the right side of the road adjacent to the garage exit, and the third camera device is used for shooting the vehicle condition on the road inside the garage exit;
step S15: carrying out deformation processing on a preset number of continuous images acquired by the three camera devices in real time through an image deformation processing unit;
step S16: processing the image processed by the image deformation processing unit through a pedestrian detection unit and judging whether a pedestrian exists in the image; extracting the features of the image and matching the features with the pedestrian model features obtained by the convolutional neural network training, if the features of the pedestrian in the image are matched with the features of the pedestrian model obtained by the convolutional neural network training, judging that the pedestrian exists in the image, and then executing the step S17; otherwise, returning to execute the step S12;
step S17: firstly, marking pedestrians in an image, then calculating and judging the moving direction of the marked pedestrians in the image through a pedestrian direction detection unit, and calculating the distance between the marked pedestrians in the image and the garage exit through a pedestrian distance detection unit;
step S18: processing the image processed by the image deformation processing unit through a vehicle detection unit and judging whether a vehicle exists in the image; extracting the features of the image, matching the extracted features with various vehicle model features obtained by training a convolutional neural network, judging that a vehicle exists in the image if the features of the vehicle in the image are matched with the vehicle model features obtained by training the convolutional neural network, and then executing step S19; otherwise, returning to execute the step S12;
step S19: firstly, marking the vehicles in the image, then calculating and judging the moving direction of the marked vehicles in the image through a vehicle direction detection unit, and identifying and judging the license plate numbers of the marked vehicles through a license plate number detection unit;
step S20: when the pedestrian detection unit detects that pedestrians are arranged on two sides of the garage outlet, the pedestrian direction detection unit detects that the pedestrians move towards the garage outlet, and the pedestrian distance detection unit detects that the pedestrians are located within a preset safety distance range, the pedestrian safety prompt triggering unit executes pedestrian safety prompt operation and pedestrian safety message pushing operation;
step S21: and (3) pedestrian safety prompting operation: the server adjusts one or more modes of brightness, flicker frequency, color, on-off state or voice broadcast of a first prompt lamp, a second prompt lamp and a third prompt lamp close to the garage exit position through the driving unit to perform safety prompt and guidance on the car owner at the garage exit position;
step S22: pedestrian safety message pushing operation: the server firstly obtains the license plate number of the current vehicle through the license plate number detection unit, and then reads the license plate number-vehicle-mounted WiFi _ ID corresponding table information stored in the storage unit so as to determine the vehicle-mounted WiFi _ ID information corresponding to the license plate number of the current vehicle; then, the server sends pedestrian data information on two sides of the garage outlet to vehicle-mounted WiFI equipment of the current vehicle through a wireless communication unit;
step S23: the vehicle-mounted WiFI equipment of the current vehicle analyzes the data information after receiving the data information sent by the wireless communication unit, and carries out pedestrian safety prompt on the vehicle owner through voice.
3. A garage exit pedestrian safety prompting system based on a convolutional neural network as claimed in claim 1, wherein when the convolutional neural network is used for training the model, 50000 pedestrian pictures with different postures are used for training the pedestrian model, 50000 license plate pictures are used for training the license plate model, and 50000 vehicle pictures with different types are respectively used for training various vehicle models.
4. A garage exit pedestrian safety warning system based on convolutional neural network as claimed in claim 1, wherein the light guide plate is rectangular in shape, and the bottom surface of the light guide plate is printed with a pattern.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011512932.6A CN112581748A (en) | 2020-12-20 | 2020-12-20 | Garage traffic-out pedestrian safety prompting system based on convolutional neural network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011512932.6A CN112581748A (en) | 2020-12-20 | 2020-12-20 | Garage traffic-out pedestrian safety prompting system based on convolutional neural network |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112581748A true CN112581748A (en) | 2021-03-30 |
Family
ID=75136326
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011512932.6A Pending CN112581748A (en) | 2020-12-20 | 2020-12-20 | Garage traffic-out pedestrian safety prompting system based on convolutional neural network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112581748A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113257008A (en) * | 2021-05-12 | 2021-08-13 | 兰州交通大学 | Pedestrian flow dynamic control system and method based on deep learning |
CN113283367A (en) * | 2021-06-08 | 2021-08-20 | 南通大学 | Safety detection method for visual blind area of underground garage in low-visibility environment |
CN114056384A (en) * | 2022-01-17 | 2022-02-18 | 矿冶科技集团有限公司 | Safety early warning system for electric locomotive and early warning method thereof |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2004349926A (en) * | 2003-05-21 | 2004-12-09 | Suzuki Motor Corp | Obstacle detection apparatus and method, and program for obstacle detection |
CN107967817A (en) * | 2017-11-17 | 2018-04-27 | 张慧 | Intelligent managing system for parking lot and method based on multi-path camera deep learning |
CN108245863A (en) * | 2018-02-05 | 2018-07-06 | 宋彦震 | Skiing blind area safety prompting system and its control method |
CN109859506A (en) * | 2019-03-26 | 2019-06-07 | 刘廷森 | Garage mouth goes out the double prompt early warning systems of vehicle and garage exit accidents early warning method |
-
2020
- 2020-12-20 CN CN202011512932.6A patent/CN112581748A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2004349926A (en) * | 2003-05-21 | 2004-12-09 | Suzuki Motor Corp | Obstacle detection apparatus and method, and program for obstacle detection |
CN107967817A (en) * | 2017-11-17 | 2018-04-27 | 张慧 | Intelligent managing system for parking lot and method based on multi-path camera deep learning |
CN108245863A (en) * | 2018-02-05 | 2018-07-06 | 宋彦震 | Skiing blind area safety prompting system and its control method |
CN109859506A (en) * | 2019-03-26 | 2019-06-07 | 刘廷森 | Garage mouth goes out the double prompt early warning systems of vehicle and garage exit accidents early warning method |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113257008A (en) * | 2021-05-12 | 2021-08-13 | 兰州交通大学 | Pedestrian flow dynamic control system and method based on deep learning |
CN113283367A (en) * | 2021-06-08 | 2021-08-20 | 南通大学 | Safety detection method for visual blind area of underground garage in low-visibility environment |
CN114056384A (en) * | 2022-01-17 | 2022-02-18 | 矿冶科技集团有限公司 | Safety early warning system for electric locomotive and early warning method thereof |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN112581748A (en) | Garage traffic-out pedestrian safety prompting system based on convolutional neural network | |
CN108645628B (en) | Automatic driving automobile test system based on road driving skills | |
CN109334566B (en) | Method, device, equipment and storage medium for providing feedback outside vehicle | |
EP1930863B1 (en) | Detecting and recognizing traffic signs | |
CN108883725A (en) | A kind of driving vehicle warning system and method | |
CN110781883B (en) | Parking space size recognition system and method based on 360-degree looking-around camera | |
CN104157143B (en) | Parking offense detection system and detection method thereof | |
JP2002083297A (en) | Object recognition method and object recognition device | |
CN102034087A (en) | Devices and methods for classifying vehicles | |
US11482012B2 (en) | Method for driving assistance and mobile device using the method | |
CN211827564U (en) | A recognition device for detecting whether vehicle opens high beam night | |
CN113055649A (en) | Tunnel intelligent video monitoring method and device, intelligent terminal and storage medium | |
CN112606831A (en) | Anti-collision warning information external interaction method and system for passenger car | |
CN105574502A (en) | Automatic detection method for violation behaviors of self-service card sender | |
CN112102609A (en) | Intelligent detection system for highway vehicle violation behaviors | |
CN112030814A (en) | Virtual barrier gate system and control method of virtual barrier gate | |
CN113808418A (en) | Road condition information display system, method, vehicle, computer device and storage medium | |
KR100743610B1 (en) | A system for photographing the car | |
EP4070231A1 (en) | Vehicle imaging station | |
CN108973841A (en) | Vehicular exterior lighting is interrupted | |
KR102252913B1 (en) | System and method of detecting signal light | |
CN113223276B (en) | Pedestrian hurdling behavior alarm method and device based on video identification | |
CN115482684A (en) | Safe driving control method and system | |
CN114582146A (en) | Traffic light remaining duration intelligent reminding method and system, storage medium and automobile | |
KR101080869B1 (en) | A System and Method for photographing the car |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210330 |
|
RJ01 | Rejection of invention patent application after publication |