CN112364820A - Deep learning-based vehicle insurance underwriting and vehicle checking picture acquisition method and system - Google Patents
Deep learning-based vehicle insurance underwriting and vehicle checking picture acquisition method and system Download PDFInfo
- Publication number
- CN112364820A CN112364820A CN202011367482.6A CN202011367482A CN112364820A CN 112364820 A CN112364820 A CN 112364820A CN 202011367482 A CN202011367482 A CN 202011367482A CN 112364820 A CN112364820 A CN 112364820A
- Authority
- CN
- China
- Prior art keywords
- vehicle
- picture
- corner
- lamp
- pictures
- 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
- 238000000034 method Methods 0.000 title claims abstract description 39
- 238000013135 deep learning Methods 0.000 title claims abstract description 14
- 240000004282 Grewia occidentalis Species 0.000 claims abstract description 82
- 238000012795 verification Methods 0.000 claims abstract description 22
- 238000001514 detection method Methods 0.000 claims description 29
- 230000002159 abnormal effect Effects 0.000 claims description 7
- 238000013528 artificial neural network Methods 0.000 abstract description 3
- 238000011156 evaluation Methods 0.000 description 5
- 239000000126 substance Substances 0.000 description 5
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000019771 cognition Effects 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000007790 scraping Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/08—Insurance
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/08—Detecting or categorising vehicles
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Business, Economics & Management (AREA)
- Accounting & Taxation (AREA)
- Multimedia (AREA)
- Evolutionary Computation (AREA)
- Finance (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Biomedical Technology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- Technology Law (AREA)
- General Business, Economics & Management (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a method and a system for acquiring car insurance underwriting car-checking pictures based on deep learning, wherein the method comprises the following steps: scanning a left front side vehicle appearance picture, a right front side vehicle appearance picture, a left rear side vehicle appearance picture or a right rear side vehicle appearance picture; carrying out vehicle attitude identification and attitude verification on image frames of a scanning video in real time; after the verification is passed, capturing a left front side vehicle four-corner picture, a right front side vehicle four-corner picture, a left rear side vehicle four-corner picture or a right rear side vehicle four-corner picture; detecting lamp components of the pictures of the four corners of the vehicle; after the lamp is detected, zooming and zooming the camera lens by taking the lamp as an anchor point, and capturing local pictures of the corresponding four-corner pictures; and uploading corresponding vehicle four-corner pictures and local pictures. The deep neural network-based classification method can obtain extremely high accuracy and recall rate.
Description
Technical Field
The invention relates to the technical field of vehicle damage assessment, in particular to a method and a system for acquiring vehicle insurance underwriting and vehicle testing images based on deep learning.
Background
Vehicle exterior parts: generally refers to the exterior components of a vehicle, such as a bumper assembly, fender assembly, door assembly, tire hub, etc., and does not include accessories such as an engine, water tank, etc., that are visible after opening the front cover.
And (4) checking the vehicle by an insurance: the vehicle participates in insurance, the vehicle owner inputs vehicle information of vehicle insurance on an insurance policy, and an insurance company sends a vehicle inspector to inspect the vehicle, and generally needs to collect vehicle four-corner pictures (including a left front side vehicle four-corner picture, a right front side vehicle four-corner picture, a left rear side vehicle four-corner picture and a right rear side vehicle four-corner picture as shown in fig. 1) for file storage in addition to comparing the vehicle identity information, so that the vehicle appearance details in the future claim settlement process are compared and used conveniently.
In the prior art, when checking the automobile by a check, four-corner pictures of the automobile appearance need to be collected, and then the damage of the automobile appearance is checked manually based on the collected four-corner pictures; the requirement is common in automobile business, such as automobile insurance underwriting, time-sharing leasing and automobile taking and returning links of automobile daily leasing, and whether the appearance of an automobile is damaged or not needs to be confirmed.
For the collection of vehicle pictures, the prior art scheme includes:
the prior patent CN108647700A discloses a multitask vehicle component recognition model, method and system based on deep learning, in which after image enhancement is performed on a remote vehicle picture, a multi-label ResNet classification network is trained; the identification system supports 3 types of identified vehicle components, namely: car lights-rearview mirrors, emblems, windshields; in addition, the system supports recognition of vehicle types and vehicle body colors.
Prior patent CN201711347241.3 discloses a self-service claims settlement method, device, equipment and computer storage medium for car insurance; the self-service claims settlement method is used for collecting the report information of the damage assessment and pushing the report information to a customer after the damage assessment is finished, so that the self-service claims settlement is finished.
The prior patent CN111652087A discloses a vehicle checking method, a device, an electronic device and a storage medium, the invention carries out image recognition and verification on the identity of a vehicle during checking, the flow is image recognition of the appearance of the vehicle, and similarity measurement is carried out by combining with a sample library type vehicle picture.
The existing system has the following problems:
1. the user mainly shoots the vehicle at four angles based on the voice prompt or the example image in the shooting process, but the cognition of the user is relatively limited, and the uploaded image is checked by a background manual or machine to know whether the uploaded image meets the shooting specification or not, so that the delay is high, and the secondary shooting influences the experience of the user;
2. after the vehicle appearance four-corner pictures are uploaded, appearance damage assessment is carried out by an underwriter, and as the vehicle appearance four-corner pictures need to clearly see the overall appearance of the vehicle, the shooting distance is usually long, so that the appearance damage (especially small-area scraping or sinking) imaging size is small, and manual examination is difficult; meanwhile, insurance companies usually compress the quality of uploaded pictures and then store the compressed pictures in a system, so that the quality of the pictures is further reduced, energy and historical experience of the underwriters are greatly consumed, and the auditing efficiency is low.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a car insurance underwriting car-checking picture acquisition method and system based on deep learning.
The invention discloses a deep learning-based vehicle insurance underwriting and vehicle checking picture acquisition method, which comprises the following steps:
scanning a left front side vehicle appearance picture, a right front side vehicle appearance picture, a left rear side vehicle appearance picture or a right rear side vehicle appearance picture;
carrying out vehicle attitude identification and attitude verification on image frames of a scanning video in real time;
after the verification is passed, capturing a left front side vehicle four-corner picture, a right front side vehicle four-corner picture, a left rear side vehicle four-corner picture or a right rear side vehicle four-corner picture;
detecting lamp components of the pictures of the four corners of the vehicle;
after the lamp is detected, zooming and zooming the camera lens by taking the lamp as an anchor point, and capturing local pictures of the corresponding four-corner pictures;
and uploading corresponding vehicle four-corner pictures and local pictures.
As a further improvement of the present invention, the attitude verification includes:
setting an attitude threshold;
when the recognition probability of the vehicle posture of the image frame is greater than the posture threshold value, the posture check is passed;
and when the recognition probability of the vehicle posture of the image frame is not greater than the posture threshold value, the posture check is not passed.
As a further refinement of the invention, the attitude threshold is 0.9.
As a further improvement of the present invention, when the attitude check is not passed, "shooting in a 45 ° direction away from the left front side of the vehicle", "shooting in a 45 ° direction away from the right front side of the vehicle", "shooting in a 45 ° direction away from the left rear side of the vehicle", or "shooting in a 45 ° direction away from the right rear side of the vehicle" is prompted.
As a further improvement of the invention, when the left front lamp of the left front vehicle four-corner picture, the right front lamp of the right front vehicle four-corner picture, the left rear lamp of the left rear vehicle four-corner picture or the right rear lamp of the right rear vehicle four-corner picture are detected, the detection is completed once at an interval of 3-8 frames.
As a further improvement of the invention, the method also comprises the following steps:
setting a detection time threshold of the lamp component;
after the lamp is detected within the detection time threshold, zooming to a camera lens by taking the lamp as an anchor point, and capturing local pictures of the corresponding four-corner pictures;
and after the lamp is not detected within the detection time threshold, the overtime exit is carried out.
As a further improvement of the present invention, the detection time threshold is 30 s.
As a further improvement of the invention, after the lamp is not detected within the detection time threshold, the vehicle four-corner picture and the abnormal state code are uploaded, and the local picture is not uploaded.
The invention also discloses a car insurance underwriting car-checking picture acquisition system based on deep learning, which comprises the following steps:
the acquisition module is used for scanning the appearance picture of the left front side vehicle, the appearance picture of the right front side vehicle, the appearance picture of the left rear side vehicle or the appearance picture of the right rear side vehicle;
the gesture recognition and verification module is used for carrying out vehicle gesture recognition and gesture verification on the image frames of the scanning video in real time;
the first capturing module is used for capturing a left front vehicle four-corner picture, a right front vehicle four-corner picture, a left rear vehicle four-corner picture or a right rear vehicle four-corner picture after the verification is passed;
the lamp component detection module is used for detecting lamp components of the pictures at the four corners of the vehicle;
the second capturing module is used for zooming and zooming the camera lens by taking the lamp as an anchor point after the lamp is detected, and capturing local pictures of the corresponding four-corner pictures;
and the uploading module is used for uploading corresponding vehicle four-corner pictures and local pictures.
As a further improvement of the present invention, the acquisition module, the gesture recognition and verification module, the first capture module, the lamp component detection module, the second capture module, and the upload module are integrated on a mobile phone.
Compared with the prior art, the invention has the beneficial effects that:
1. the deep neural network-based classification method can obtain extremely high accuracy and recall rate;
2. the method can check, feed back and adjust the vehicle posture of the shot picture in real time, is convenient for a user to submit the standard and standard four-corner picture, avoids high delay of background manual check, and has good user experience.
3. The method carries out real-time part detection on the four-corner picture with standard specification, carries out lens zooming based on the lamp part anchor point, then triggers and captures the vehicle appearance local picture under the view angle, is convenient for background manual evaluation of damage details or an intelligent damage detection system to carry out damage verification, and is more reasonable in damage details compared with the method that the four-corner picture is directly used.
Drawings
FIG. 1 is a photograph of the four corners of a vehicle;
fig. 2 is a flowchart of an embodiment of the invention, which discloses a deep learning-based car insurance underwriting and car checking image acquisition method.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, are within the scope of the present invention.
The invention is described in further detail below with reference to the attached drawing figures:
as shown in fig. 2, the invention provides a deep learning-based vehicle insurance underwriting and vehicle checking image acquisition method, which comprises the following steps:
step 1, scanning a left front side vehicle appearance picture, a right front side vehicle appearance picture, a left rear side vehicle appearance picture or a right rear side vehicle appearance picture;
wherein the content of the first and second substances,
the step 1 is realized based on a mobile phone camera;
one implementation method is as follows: opening a camera of the program, and directly scanning the vehicle appearance picture from any direction of four corners of the vehicle;
the other realization method comprises the following steps: firstly, selecting a vehicle four-corner picture to be captured on a mobile phone display screen, then opening a camera of the program, and scanning a vehicle appearance picture from the position of the selected vehicle four-corner picture; for example, a four-corner picture of a left front vehicle to be captured is selected on a display screen of a mobile phone, then a camera of the program is opened, and an appearance picture of the left front vehicle is scanned;
step 2, carrying out vehicle attitude identification and attitude verification on the image frames of the scanning video in real time;
wherein the content of the first and second substances,
the step 1 is realized based on an internal processor of the mobile phone;
the specific implementation method comprises the following steps:
carrying out vehicle attitude identification on image frames of the scanning video frame by frame to obtain the identification probability of the vehicle attitude of the image frames; when the recognition probability of the vehicle posture of the image frame is greater than a posture threshold value (preferably 0.9), the posture check is passed; and when the recognition probability of the vehicle posture of the image frame is not greater than the posture threshold value, the posture check is not passed.
Further, when the posture check is not passed, prompting "shoot away from the left front side of the vehicle by 45 degrees", "shoot away from the right front side of the vehicle by 45 degrees", "shoot away from the left rear side of the vehicle by 45 degrees" or "shoot away from the right rear side of the vehicle by 45 degrees"; for example, when the vehicle posture check of the vehicle appearance picture of the left front side fails, the mobile phone display screen prompts the adjustment of the mobile phone posture and distance, namely prompting the shooting in the direction of 45 degrees away from the left front side of the vehicle.
Step 3, correspondingly capturing a left front vehicle four-corner picture, a right front vehicle four-corner picture, a left rear vehicle four-corner picture or a right rear vehicle four-corner picture after the current vehicle posture is verified;
for example, after the vehicle posture check on the left front vehicle appearance picture passes, the mobile phone captures a left front vehicle four-corner picture as the 1 st picture of the four-corner pictures.
Step 4, detecting lamp components of the pictures at four corners of the vehicle;
wherein the content of the first and second substances,
detecting left front side lamp parts for the picture of the four corners of the left front side vehicle, or
Detecting the right front lamp part of the four-corner picture of the right front vehicle, or
Detecting left rear lamp parts of the four-corner picture of the left rear vehicle, or
And detecting the right rear lamp part of the four-corner picture of the right rear vehicle.
Step 5, after the lamp is detected, zooming the camera lens by taking the lamp as an anchor point, and capturing local pictures of the corresponding four-corner pictures;
wherein the content of the first and second substances,
the local picture is a local picture taking the lamp as the center of the anchor point and comprises one quarter of vehicle appearance parts; for example, the left front partial picture should contain left front vehicle appearance components.
Further, when a left front lamp of the left front vehicle four-corner picture, a right front lamp of the right front vehicle four-corner picture, a left rear lamp of the left rear vehicle four-corner picture or a right rear lamp of the right rear vehicle four-corner picture are detected, the detection is completed once at intervals of 3-8 frames; preferably once every 5 frames.
Further, after the lamp is not detected within the detection time threshold, the operation is exited overtime (the overtime time is usually set to 30s), the local picture is not uploaded, and only one abnormal state code is submitted.
6, uploading corresponding vehicle four-corner pictures and local pictures;
wherein the content of the first and second substances,
and uploading the pictures (4) at the four corners of the vehicle and the corresponding local pictures (4) to a back-end system for evaluation of damage details by an underwriter or an intelligent damage detection system. The vehicle four-corner pictures and the local pictures can be uploaded simultaneously or in real time.
Further, after the lamp is not detected within the detection time threshold, the images of the four corners of the vehicle and the abnormal state codes are uploaded, and the local images are not uploaded. For example, if no lamp is detected in the picture of the four corners of the left front vehicle within a preset time, lamps are detected by other triangles; then the four corner pictures (4) of the vehicle, the abnormal state code at the left front side and the corresponding partial pictures (3, right front, left back and right back) are displayed.
Further, a deep learning network mobilenet is adopted for recognizing the vehicle posture at the mobile phone end, and the category comprises four visual angles; the detection model of the mobile phone end vehicle lamp component adopts a deep learning network Frcnn structure, and only detects the vehicle lamp, including a headlamp and a tail lamp component.
The invention also discloses a car insurance underwriting car-checking picture acquisition system based on deep learning, which comprises the following steps:
the acquisition module is used for realizing the step 1;
the gesture recognition and verification module is used for realizing the step 2;
a first capturing module, configured to implement step 3;
a lamp component detection module for implementing the step 4;
a second capturing module for implementing the step 5;
and the uploading module is used for realizing the step 6.
Furthermore, the acquisition module, the gesture recognition and verification module, the first capture module, the lamp component detection module, the second capture module and the uploading module are integrated on the mobile phone.
Example (b):
1. left front vehicle four corner picture collection
A customer holds a mobile phone to scan the appearance picture of the left front vehicle, and a mobile phone end can perform vehicle posture recognition on the image frame of a scanned video in real time; the recognition probability of the vehicle posture of the image frame is greater than 0.9, the posture is verified to be passed, and the mobile phone captures a four-corner picture of the left front vehicle;
after the left front vehicle picture is uploaded, detecting a left front lamp component on the left front vehicle picture; when the lamp is detected, the lamp is used as an anchor point, the zoom lens captures a local picture of the left front side vehicle after the view field image is stable and uploads the local picture, and the picture is used for damage detail evaluation of an underwriter or an intelligent damage detection system.
2. Right front side vehicle four corner picture collection
A customer holds a mobile phone to scan the appearance picture of the vehicle at the right front side, and a mobile phone end can perform vehicle posture recognition on the image frame of a scanned video in real time; the recognition probability of the vehicle posture of the image frame is less than 0.9, after the posture check is not passed, prompting to shoot in a direction of 45 degrees away from the right front side of the vehicle until the recognition probability of the vehicle posture of the image frame is more than 0.9; the gesture is checked to be passed, and the mobile phone captures the four-corner picture of the vehicle at the front right side;
after the right front side vehicle picture is uploaded, detecting a right front side lamp component on the right front side vehicle picture; when the lamp is detected, the lamp is used as an anchor point, the zoom lens captures a local picture of the left front side vehicle after the view field image is stable and uploads the local picture, and the picture is used for damage detail evaluation of an underwriter or an intelligent damage detection system.
3. Left rear vehicle four corner picture collection
A customer holds a mobile phone to scan the appearance picture of the left rear vehicle, and a mobile phone end can perform vehicle posture recognition on the image frame of a scanned video in real time; the recognition probability of the vehicle posture of the image frame is greater than 0.9, the posture is verified to be passed, and the mobile phone captures the four-corner picture of the left rear vehicle;
after the left rear vehicle picture is uploaded, detecting a left rear lamp component on the left rear vehicle picture; and when the left rear lamp is not detected within 30s, the local picture is not uploaded, and only one abnormal state code is submitted.
4. Right rear vehicle four corner picture collection
A customer holds a mobile phone to scan the appearance picture of the right rear vehicle, and the mobile phone end can perform vehicle posture recognition on the image frame of a scanned video in real time; the recognition probability of the vehicle posture of the image frame is less than 0.9, after the posture check is not passed, prompting to shoot in a direction of 45 degrees away from the right front side of the vehicle until the recognition probability of the vehicle posture of the image frame is more than 0.9; the gesture is checked to be passed, and the mobile phone captures the four-corner pictures of the right rear vehicle;
after the right rear vehicle picture is uploaded, detecting a right rear lamp component on the right rear vehicle picture; and when the right rear side lamp is not detected within 30s, the local picture is not uploaded, and only one abnormal state code is submitted.
The method is a self-service acquisition process of the four-corner pictures of the underwriting inspection vehicle, and has the following advantages:
1. the deep neural network-based classification method can obtain extremely high accuracy and recall rate;
2. the method can check, feed back and adjust the vehicle posture of the shot picture in real time, is convenient for a user to submit the standard and standard four-corner picture, avoids high delay of background manual check, and has good user experience.
3. The method carries out real-time part detection on the four-corner picture with standard specification, carries out lens zooming based on the lamp part anchor point, then triggers and captures the vehicle appearance local picture under the view angle, is convenient for background manual evaluation of damage details or an intelligent damage detection system to carry out damage verification, and is more reasonable in damage details compared with the method that the four-corner picture is directly used.
The above is only a preferred embodiment of the present invention, and is not intended to limit the present invention, and various modifications and changes will occur to those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. A car insurance underwriting and car checking picture acquisition method based on deep learning is characterized by comprising the following steps:
scanning a left front side vehicle appearance picture, a right front side vehicle appearance picture, a left rear side vehicle appearance picture or a right rear side vehicle appearance picture;
carrying out vehicle attitude identification and attitude verification on image frames of a scanning video in real time;
after the verification is passed, capturing a left front side vehicle four-corner picture, a right front side vehicle four-corner picture, a left rear side vehicle four-corner picture or a right rear side vehicle four-corner picture;
detecting lamp components of the pictures of the four corners of the vehicle;
after the lamp is detected, zooming and zooming the camera lens by taking the lamp as an anchor point, and capturing local pictures of the corresponding four-corner pictures;
and uploading corresponding vehicle four-corner pictures and local pictures.
2. The acquisition method as set forth in claim 1, wherein the pose verification comprises:
setting an attitude threshold;
when the recognition probability of the vehicle posture of the image frame is greater than the posture threshold value, the posture check is passed;
and when the recognition probability of the vehicle posture of the image frame is not greater than the posture threshold value, the posture check is not passed.
3. The acquisition method as set forth in claim 2, wherein the pose threshold is 0.9.
4. The acquisition method according to claim 2, wherein when the attitude check is not passed, the "photographing in a 45 ° direction away from the left front side of the vehicle", "photographing in a 45 ° direction away from the right front side of the vehicle", "photographing in a 45 ° direction away from the left rear side of the vehicle", or "photographing in a 45 ° direction away from the right rear side of the vehicle" is prompted.
5. The acquisition method according to claim 1, wherein the detection of the left front lamp of the left front vehicle corner picture, the right front lamp of the right front vehicle corner picture, the left rear lamp of the left rear vehicle corner picture or the right rear lamp of the right rear vehicle corner picture is completed at intervals of 3-8 frames.
6. The acquisition method as set forth in claim 1, further comprising:
setting a detection time threshold of the lamp component;
after the lamp is detected within the detection time threshold, zooming to a camera lens by taking the lamp as an anchor point, and capturing local pictures of the corresponding four-corner pictures;
and after the lamp is not detected within the detection time threshold, the overtime exit is carried out.
7. The acquisition method according to claim 6, characterized in that said detection time threshold is 30 s.
8. The acquisition method according to claim 6, wherein after the lamp is not detected within the detection time threshold, the vehicle four-corner picture and the abnormal state code are uploaded, and the local picture is not uploaded.
9. An acquisition system for implementing an acquisition method according to any one of claims 1 to 8, comprising:
the acquisition module is used for scanning the appearance picture of the left front side vehicle, the appearance picture of the right front side vehicle, the appearance picture of the left rear side vehicle or the appearance picture of the right rear side vehicle;
the gesture recognition and verification module is used for carrying out vehicle gesture recognition and gesture verification on the image frames of the scanning video in real time;
the first capturing module is used for capturing a left front vehicle four-corner picture, a right front vehicle four-corner picture, a left rear vehicle four-corner picture or a right rear vehicle four-corner picture after the verification is passed;
the lamp component detection module is used for detecting lamp components of the pictures at the four corners of the vehicle;
the second capturing module is used for zooming and zooming the camera lens by taking the lamp as an anchor point after the lamp is detected, and capturing local pictures of the corresponding four-corner pictures;
and the uploading module is used for uploading corresponding vehicle four-corner pictures and local pictures.
10. The acquisition system of claim 9 wherein the acquisition module, gesture recognition and verification module, first capture module, light fixture component detection module, second capture module, and upload module are integrated on a cell phone.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011367482.6A CN112364820A (en) | 2020-11-27 | 2020-11-27 | Deep learning-based vehicle insurance underwriting and vehicle checking picture acquisition method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011367482.6A CN112364820A (en) | 2020-11-27 | 2020-11-27 | Deep learning-based vehicle insurance underwriting and vehicle checking picture acquisition method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112364820A true CN112364820A (en) | 2021-02-12 |
Family
ID=74536239
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011367482.6A Pending CN112364820A (en) | 2020-11-27 | 2020-11-27 | Deep learning-based vehicle insurance underwriting and vehicle checking picture acquisition method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112364820A (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107204114A (en) * | 2016-03-18 | 2017-09-26 | 中兴通讯股份有限公司 | A kind of recognition methods of vehicle abnormality behavior and device |
| US9886771B1 (en) * | 2016-05-20 | 2018-02-06 | Ccc Information Services Inc. | Heat map of vehicle damage |
| CN108921068A (en) * | 2018-06-22 | 2018-11-30 | 深源恒际科技有限公司 | A kind of automatic damage identification method of automobile appearance based on deep neural network and system |
| CN109325488A (en) * | 2018-08-31 | 2019-02-12 | 阿里巴巴集团控股有限公司 | For assisting the method, device and equipment of car damage identification image taking |
| CN109344819A (en) * | 2018-12-13 | 2019-02-15 | 深源恒际科技有限公司 | Vehicle damage recognition methods based on deep learning |
| CN110910628A (en) * | 2019-12-02 | 2020-03-24 | 支付宝(杭州)信息技术有限公司 | Interactive processing method and device for vehicle damage image shooting and electronic equipment |
| CN111680596A (en) * | 2020-05-29 | 2020-09-18 | 北京百度网讯科技有限公司 | Positioning truth value verification method, device, equipment and medium based on deep learning |
-
2020
- 2020-11-27 CN CN202011367482.6A patent/CN112364820A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107204114A (en) * | 2016-03-18 | 2017-09-26 | 中兴通讯股份有限公司 | A kind of recognition methods of vehicle abnormality behavior and device |
| US9886771B1 (en) * | 2016-05-20 | 2018-02-06 | Ccc Information Services Inc. | Heat map of vehicle damage |
| CN108921068A (en) * | 2018-06-22 | 2018-11-30 | 深源恒际科技有限公司 | A kind of automatic damage identification method of automobile appearance based on deep neural network and system |
| CN109325488A (en) * | 2018-08-31 | 2019-02-12 | 阿里巴巴集团控股有限公司 | For assisting the method, device and equipment of car damage identification image taking |
| CN109344819A (en) * | 2018-12-13 | 2019-02-15 | 深源恒际科技有限公司 | Vehicle damage recognition methods based on deep learning |
| CN110910628A (en) * | 2019-12-02 | 2020-03-24 | 支付宝(杭州)信息技术有限公司 | Interactive processing method and device for vehicle damage image shooting and electronic equipment |
| CN111680596A (en) * | 2020-05-29 | 2020-09-18 | 北京百度网讯科技有限公司 | Positioning truth value verification method, device, equipment and medium based on deep learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP6461406B1 (en) | Apparatus, system and method for improved face detection and recognition in a vehicle inspection security system | |
| CN112712037A (en) | Vehicle-mounted environment sensing method and system based on panoramic image and target detection | |
| CN109740547A (en) | A kind of image processing method, equipment and computer readable storage medium | |
| CN105654733A (en) | Front and back vehicle license plate recognition method and device based on video detection | |
| CN113038018A (en) | Method and device for assisting user in shooting vehicle video | |
| CN106585530A (en) | Vehicle-mounted snapshooting device and control method thereof | |
| CN111523352A (en) | Method for intelligently and rapidly identifying illegal modified vehicle and monitoring system thereof | |
| CN107527095A (en) | A kind of vehicle detection image-pickup method and system | |
| WO2020140611A1 (en) | Vin code identification method based on image processing, device and medium | |
| CN115810134B (en) | Image acquisition quality inspection method, system and device for vehicle insurance anti-fraud | |
| CN112364820A (en) | Deep learning-based vehicle insurance underwriting and vehicle checking picture acquisition method and system | |
| WO2020234089A1 (en) | Automatic damage detection for cars | |
| JP2006344154A (en) | Vehicle discriminating device and method | |
| KR102428930B1 (en) | Method and system for payment of car wash charge based on car identification | |
| CN115457427A (en) | Refueling process monitoring method and device, electronic equipment and storage medium | |
| CN108737733A (en) | Information cuing method and device, electronic equipment, computer readable storage medium | |
| CN114373303A (en) | Mobile modularized public security intelligent inspection station system | |
| CN112492105B (en) | Video-based vehicle appearance part self-service damage assessment acquisition method and system | |
| CN115205990A (en) | Parking lot charging system based on face recognition | |
| CN107194385A (en) | A kind of intelligent vehicle license plate recognition system | |
| KR100855590B1 (en) | System and method for multi car recognition of multi lens and multi view point | |
| CN117455466B (en) | Method and system for remote evaluation of automobile | |
| CN110826473A (en) | Neural network-based automatic insulator image identification method | |
| CN114170583B (en) | Vehicle detecting system based on convolutional neural network | |
| CN117671328A (en) | Vehicle damage assessment method, device, computer equipment and storage medium |
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 |