A method and system for automatic assessment of asset loss before and after an incident

Abstract

The application provides an accident before and after asset loss automatic evaluation method and system, and the method comprises the following steps: acquiring panoramic pictures of each scanning point by a panoramic camera, and generating a scene three-dimensional model based on the panoramic pictures; acquiring asset information of each object in the scene three-dimensional model to form an asset information list; after an accident occurs, repeating the above steps to acquire an asset information list after the accident; comparing the asset information of the asset information list before and after the accident to acquire asset difference information of each object, and calculating the asset loss. The application only needs to use a specific device to scan on the scene, generates a three-dimensional digital twin space, and compares the asset loss through AI, without manual loss checking, and also does not need to take a large number of pictures on the scene, avoiding missing part of the scene, and because the digital twin space of the scene is completely copied, the accuracy of the asset loss comparison is greatly improved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to an automatic method and system for assessing asset losses before and after an accident. Background Technology

[0002] When an accident occurs in a building (such as a fire or explosion), it is necessary to compare the losses of the assets within the building. Currently, this is typically done manually or through computer vision. Manual verification is very costly in terms of manpower and time; while assessment requires taking numerous photos of the site, which are only two-dimensional images and may have limitations, resulting in lower accuracy for loss comparisons using computer vision. Summary of the Invention

[0003] To overcome the shortcomings of existing technologies, this invention provides an automatic asset loss assessment method and system before and after an accident, which solves the problems of high labor costs, large workload, and poor accuracy of existing asset loss assessment methods.

[0004] The technical solution adopted by this invention to solve its technical problem is:

[0005] This invention first provides an automatic asset loss assessment method before and after an accident, comprising the following steps:

[0006] S1. Obtain panoramic images of each scanning point using a panoramic camera, and generate a 3D model of the scene based on the panoramic images;

[0007] S2. Extract and analyze the scene 3D model and panoramic image, obtain the item asset information of each item in the scene 3D model, classify, record and save it to form an asset information list;

[0008] S3. After the accident, repeat steps S1 and S2 to obtain a list of asset information after the accident.

[0009] S4. Compare the item asset information in the asset information list before and after the accident, obtain the asset difference information for each item, and calculate the asset loss based on the asset difference information.

[0010] As a further preferred embodiment, the item asset information in step S2 includes category information, location information, and digital characteristic information;

[0011] As a further preferred embodiment, the method for obtaining the item asset information in step S2 is as follows:

[0012] S21. Based on deep learning, two-dimensional image target detection technology and instance segmentation technology are used to analyze the panoramic image information obtained from each scanning point, so as to identify each item in the panoramic image and obtain the category information, location information and digital feature information of each item.

[0013] S22. Obtain the position information of each item in the scene 3D model, and determine whether the same item appears repeatedly in multiple panoramic images based on the correspondence between the item position information in the scene 3D model and the item position information in the panoramic image;

[0014] S23. If the same item does not appear repeatedly in the panoramic image, the digital feature information of the item obtained from the corresponding panoramic image shall be recorded and saved as one of the item asset information of the item.

[0015] If the same item appears repeatedly in multiple panoramic images, the digital feature information of the item obtained from the multiple panoramic images will be fused together, and the fused feature information will be recorded and saved as one of the item asset information of the item.

[0016] As a further preferred implementation, the specific operation of fusing the digital feature information of the item obtained from multiple panoramic images in step S23 is as follows:

[0017] S231. When the same item is repeatedly identified in multiple panoramic images, the digital feature information of the item obtained in the multiple panoramic images are TV0, TV1, ..., TV1, respectively. n The corresponding discrimination probabilities are TR0, TR1, ..., TR1, respectively. n ;

[0018] S232. Fuse the digital feature information of the item according to the following formula:

[0019]

[0020] Wherein, PV represents the fused digital feature information of the item.

[0021] As a further preferred implementation scheme, step S4 is specifically performed as follows:

[0022] S41. For each item C in the pre-accident asset information list. 0i Search the asset information list after the accident for items of the same category C. 1i ;

[0023] S42. If the same item category C is not found in the asset information list after the accident. 1i Then determine item C 0iThe loss rate is 100%;

[0024] If the same item category C is found in the asset information list after the accident... 1i Then calculate the percentage loss of digital feature information of the item before and after the accident, and use this percentage loss of digital feature information as the value of item C. 0i The loss rate;

[0025] S43. Repeat steps S41 and S42 to obtain the loss rate of all items, and calculate the asset loss based on the obtained item loss rate.

[0026] As a further preferred embodiment, the formula for calculating the percentage loss of digital feature information in step S42 is:

[0027]

[0028] Among them, PVO l PVO is the digital characteristic information of item i before the accident. i The (l+1)th vector in the array, PV1 l PV1 is the digital feature information of item i after the accident. i The (l+1)th vector in the array.

[0029] As a further preferred embodiment, the formula for calculating the percentage loss of digital feature information in step S42 is:

[0030]

[0031] Among them, PVO l Digital characteristic information PV0 of item i before the accident. i The (l+1)th vector in the array, PV1 l PV1 is the digital feature information of item i after the accident. i The (l+1)th vector in the array.

[0032] This invention also provides an automatic asset loss assessment system before and after an accident, comprising:

[0033] The 3D model building module is used to acquire panoramic images of each scanning point through a panoramic camera and generate a 3D model of the scene based on the panoramic images.

[0034] The asset information generation module is used to extract and analyze the scene 3D model, obtain the item asset information of each item in the scene 3D model, classify, record and save it to form an asset information list;

[0035] The asset information comparison module is used to compare the item asset information in the asset information list before and after the accident, and to obtain the asset difference information of each item.

[0036] The asset loss calculation module is used to calculate asset losses based on the asset difference information.

[0037] As a further preferred embodiment, the asset information generation module includes a feature fusion module, which is used to fuse the digital feature information of the same item obtained from multiple panoramic images.

[0038] The positive effects of this invention: This invention proposes an automatic asset loss assessment method and system before and after an accident. It only requires scanning the site with specific equipment to generate a three-dimensional digital twin space, and then comparing asset losses through AI. No manual loss verification is required, and it also does not require taking a large number of pictures of the site, thus avoiding the omission of some parts of the site. At the same time, because the digital twin space completely replicates the actual scene, the accuracy of asset loss comparison is greatly improved. Attached Figure Description

[0039] Figure 1 This is a flowchart of the automatic asset loss assessment method before and after an accident as described in this invention. Detailed Implementation

[0040] The preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0041] Reference Figure 1 A preferred embodiment of the present invention provides an automatic asset loss assessment method before and after an accident, comprising the following steps:

[0042] S1. Obtain panoramic images of each scanning point using a panoramic camera (the number of scanning points depends on the complexity of the actual shooting scene; if a cluttered environment is encountered, more points can be photographed), and generate a 3D model of the scene based on the panoramic images.

[0043] S2. Extract and analyze the scene 3D model and panoramic image, obtain the item asset information of each item in the scene 3D model, classify, record and save it to form an asset information list;

[0044] The item asset information includes category information, location information, and digital feature information (i.e., a mathematical expression of item features, generally in the form of a high-dimensional feature vector).

[0045] The method for obtaining the item asset information includes the following steps:

[0046] S21. Based on deep learning, two-dimensional image target detection technology and instance segmentation technology are used to analyze the panoramic image information obtained from each scanning point, so as to identify each item in the panoramic image and obtain the category information, location information and digital feature information of each item.

[0047] S22. Obtain the position information of each item in the scene 3D model. Based on the correspondence between the position information of the items in the scene 3D model and the position information of the items in the panoramic image, determine whether the same item appears repeatedly in multiple panoramic images (if the position relationship of an item in the 3D model corresponds to the position relationship of an item in multiple panoramic images, it means that the item appears in multiple panoramic images at the same time, that is, the item is captured in multiple panoramic images from different shooting points).

[0048] S23. If the same item does not appear repeatedly in the panoramic image, the digital feature information of the item obtained from the corresponding panoramic image shall be recorded and saved as the item asset information of the item.

[0049] If the same item appears repeatedly in multiple panoramic images, the digital feature information of the item obtained from the multiple panoramic images will be fused (that is, for the same item that appears repeatedly in multiple panoramic images, the feature information obtained from different perspectives can be fused according to a certain calculation method to obtain a more accurate overall mathematical feature expression of the item), and the fused digital feature information will be recorded and saved as the item asset information of the item.

[0050] The specific operation of fusing the feature information of the item obtained from multiple panoramic images is as follows:

[0051] S231. When the same item is repeatedly identified in multiple panoramic images, the digital feature information obtained from the multiple identifications of the item in the multiple panoramic images (assuming the number of identifications is n+1) are TV0, TV1, ..., TV2. n The corresponding discrimination probabilities are TR0, TR1, ..., TR1, respectively. n ;

[0052] S232. Fuse the digital feature information of the item according to the following formula:

[0053]

[0054] Wherein, PV represents the fused feature information of the item, that is, PV is the mathematical description of the fused digital feature information of the corresponding object. Generally, it is represented as a high-dimensional vector (assuming it is an m+1 dimensional vector), i.e.

[0055] PV = (PV0, PV1, ..., PV)m );

[0056] ω j It is the weight value when different features are fused. It can be expressed as the proportion of the corresponding item's discrimination probability in the sum of the total discrimination probabilities under the current viewpoint. Its expression is:

[0057]

[0058] For feature values ​​with high discrimination probabilities, their influence on the final feature fusion is also greater.

[0059] Furthermore, the asset information list is generated in the following way:

[0060] The category information C0 of an item i in the house. i Its digital feature information PV0 i By establishing a one-to-one correspondence and processing each item in the house separately, individual category information and digital characteristic information can be generated. This information, along with item C0... i Valuation information PE i By storing these digitally (e.g., in databases, Excel spreadsheets, asset management software, etc.), a list of all items in a house before the accident and their digital characteristics can be created. The method for generating a list of assets after subsequent accidents is the same.

[0061] S3. After the accident, repeat steps S1 and S2 to obtain a list of asset information after the accident (matching the category information C1i of an item i in the house with its digital characteristic information PV1). i By establishing a one-to-one correspondence and processing each item in the house after the accident separately, individual category information and digital characteristic information can be generated.

[0062] S4. Compare the item asset information in the asset information list before and after the accident, obtain the asset difference information for each item, and calculate the asset loss based on the asset difference information; this step S4 specifically includes the following steps:

[0063] S41. For each item C in the pre-accident asset information list. 0i Search the asset information list after the accident for items of the same category C. 1i ;

[0064] S42. If the same item category C is not found in the asset information list after the accident. 1i Then determine item C 0i The loss rate is 100%;

[0065] If the same item category C is found in the asset information list after the accident... 1i Then calculate the percentage loss of digital feature information of the item before and after the accident, and use this percentage loss of digital feature information as the value of item C. 0i The loss rate;

[0066] S43. Repeat steps S41 and S42 to obtain the loss rate of all items, and calculate the asset loss based on the obtained item loss rate.

[0067] One possible calculation method is to calculate the ratio of the differences in feature information over Euclidean distance, i.e., the formula for calculating the percentage loss of digital feature information in step S42 is:

[0068]

[0069] Alternatively, another possible calculation method is to calculate the ratio of the differences in feature information along the cosine distance, i.e., the formula for calculating the percentage loss of digital feature information in step S42 is:

[0070]

[0071] Among them, PVO l PVO is the digital characteristic information of item i before the accident. i The (l+1)th vector in the array, PV1 l PV1 is the digital feature information of item i after the accident. i The (l+1)th vector in the array.

[0072] Based on the pre-accident valuation information of every item in the house, PE i And the loss ratio of the item L i By defining certain rules, the estimated loss of an item can be easily and automatically estimated. Based on this, the estimated loss of each item in the room before the accident, as well as the total estimated loss, can be automatically provided. This invention can be applied to specific industries, including but not limited to insurance claims and police investigations.

[0073] This embodiment also provides an automatic asset loss assessment system before and after an accident, including:

[0074] The 3D model building module is used to acquire panoramic images of each scanning point through a panoramic camera and generate a 3D model of the scene based on the panoramic images.

[0075] The asset information generation module is used to extract and analyze the scene 3D model, obtain the item asset information of each item in the scene 3D model, classify, record and save it to form an asset information list;

[0076] The asset information comparison module is used to compare the asset information of items in the asset information list before and after the accident to obtain the asset difference information of each item; it also includes a feature fusion module, which is used to fuse the digital feature information of the same item obtained from multiple panoramic images.

[0077] The asset loss calculation module is used to calculate asset losses based on the asset difference information.

[0078] The above description is only a preferred embodiment of the present invention. It should be understood that the above description of the embodiments is only for the purpose of helping to understand the method and core idea of ​​the present invention, and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, etc. made within the idea and principle of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for automatically assessing asset losses before and after an accident, characterized in that, Includes the following steps: S1. Acquire panoramic images of each scanning point using a panoramic camera, and generate a 3D model of the scene based on the panoramic images; S2. Extract and analyze the 3D model and panoramic images of the scene, obtain the asset information of each item in the 3D model of the scene, classify, record and save it to form an asset information list; S3. After the accident, repeat steps S1 and S2 to obtain a list of asset information after the accident. S4. Compare the item asset information in the asset information list before and after the accident, obtain the asset difference information for each item, and calculate the asset loss based on the asset difference information; The method for obtaining the item asset information in step S2 is as follows: S21. Based on deep learning, two-dimensional image target detection technology and instance segmentation technology are used to analyze the panoramic image information obtained from each scanning point, so as to identify each item in the panoramic image and obtain the category information, location information and digital feature information of each item. S22. Obtain the position information of each item in the scene 3D model, and determine whether the same item appears repeatedly in multiple panoramic images based on the correspondence between the item position information in the scene 3D model and the item position information in the panoramic image; S23. If the same item does not appear repeatedly in the panoramic image, the digital feature information of the item obtained from the corresponding panoramic image shall be recorded and saved as one of the item asset information of the item. If the same item appears repeatedly in multiple panoramic images, the digital feature information of the item obtained from the multiple panoramic images will be fused together, and the fused digital feature information will be recorded and saved as one of the item asset information of the item. The specific operation of fusing the digital feature information of the item obtained from multiple panoramic images in step S23 is as follows: S231. When the same item is repeatedly identified in multiple panoramic images, the digital feature information of the item obtained in the multiple panoramic images are TV0, TV1, ..., TV1, respectively. n The corresponding discrimination probabilities are TR0, TR1, ..., TR1, respectively. n ; S232. Fuse the digital feature information of the item according to the following formula: in, For the fusion of digital feature information of this item, The specific operation of step S4 is as follows: S41. For each item C in the pre-accident asset information list. 0i Search the asset information list after the accident for items of the same category C. 1i ; S42. If the same item category C is not found in the asset information list after the accident. 1i Then determine item C 0i The loss rate was 100%; If the same item category C is found in the asset information list after the accident... 1i Then calculate the percentage loss of digital feature information of the item before and after the accident, and use this percentage loss of digital feature information as the value of item C. 0i The loss rate; S43. Repeat steps S41 and S42 to obtain the loss rate of all items, and calculate the asset loss based on the obtained item loss rate; The formula for calculating the percentage loss of digital feature information in step S42 is as follows: in, Digital characteristic information of item i before the accident. The first in +1 vector, The numerical characteristic information of item i after the accident. The first in +1 vector; The formula for calculating the percentage loss of digital feature information in step S42 is as follows: in, Digital characteristic information of item i before the accident. The first in +1 vector, The numerical characteristic information of item i after the accident. The first in +1 vectors.

2. The automatic asset loss assessment method before and after an accident according to claim 1, characterized in that: The item asset information mentioned in step S2 includes category information, location information, and digital characteristic information.

3. An automatic asset loss assessment system before and after an accident, characterized in that, include: The 3D model building module is used to acquire panoramic images of each scanning point through a panoramic camera and generate a 3D model of the scene based on the panoramic images. The asset information generation module is used to extract and analyze the scene 3D model, obtain the item asset information of each item in the scene 3D model, classify, record and save it to form an asset information list; the asset information generation module includes a feature fusion module, which is used to fuse the digital feature information of the same item obtained from multiple panoramic images; The method for obtaining the item asset information is as follows: Deep learning-based 2D image target detection and instance segmentation techniques are used to analyze panoramic image information obtained from each scanning point in order to identify each item in the panoramic image and obtain the category information, location information, and digital feature information of each item. Obtain the position information of each item in the scene 3D model, and determine whether the same item appears repeatedly in multiple panoramic images based on the correspondence between the item position information in the scene 3D model and the item position information in the panoramic image; If the same item does not appear repeatedly in the panoramic image, the digital feature information of the item obtained from the corresponding panoramic image will be recorded and saved as one of the item asset information of the item. If the same item appears repeatedly in multiple panoramic images, the digital feature information of the item obtained from the multiple panoramic images will be fused, and the fused digital feature information will be recorded and saved as one of the item asset information. The specific operation for fusing the digital feature information of the item obtained from multiple panoramic images is as follows: When the same item is repeatedly identified in multiple panoramic images, the digital feature information of the item obtained in the multiple panoramic images are TV0, TV1, ..., TV1, respectively. n The corresponding discrimination probabilities are TR0, TR1, ..., TR1, respectively. n ; The digital feature information of the item is fused using the following formula: in, For the fusion of digital feature information of this item, The asset information comparison module is used to compare the item asset information in the asset information list before and after the accident, and to obtain the asset difference information of each item. The asset loss calculation module is used to calculate asset losses based on the asset difference information. The specific operation is as follows: S41. For each item C in the pre-accident asset information list. 0i Search the asset information list after the accident for items of the same category C. 1i ; S42. If the same item category C is not found in the asset information list after the accident. 1i Then determine item C 0i The loss rate was 100%; If the same item category C is found in the asset information list after the accident... 1i Then calculate the percentage loss of digital feature information of the item before and after the accident, and use this percentage loss of digital feature information as the value of item C. 0i The loss rate; S43. Repeat steps S41 and S42 to obtain the loss rate of all items, and calculate the asset loss based on the obtained item loss rate; The formula for calculating the percentage loss of digital feature information in step S42 is as follows: in, Digital characteristic information of item i before the accident. The first in +1 vector, The numerical characteristic information of item i after the accident. The first in +1 vector; The formula for calculating the percentage loss of digital feature information in step S42 is as follows: in, Digital characteristic information of item i before the accident. The first in +1 vector, The numerical characteristic information of item i after the accident. The first in +1 vectors.

Images