Systems and methods for dynamic real-time analysis from multi-modal data fusion for contextual risk identification
Disclosed is a system and method to automatically identify property-related risks through the use of computer vision, sensors, and/or building information models (BIMs). The ability to automatically identify a variety of hazards helps mitigate the associated risks, and thus reduces the number of accidents or fatalities that would otherwise occur. In some embodiments, a “risk map” can be generated by mapping the identified risks for a given property.
1. A computing device, comprising:
a processor; and
a memory storing instructions that, when executed by the processor, cause the computing device to perform operations comprising:
capturing, by a camera, an image of an object at a site location;
detecting the object in the image by executing a Convolutional Neural Network (CNN) trained to detect the object in the image using a plurality of convolution layers to extract aspects of the image in a progressively hierarchical fashion by combining features from a previous convolution layer with a next convolution layer through a weighted non-linear function;
identifying, using edge analysis, an object type for the object by determining an object location of the object in the image and matching the object location to a model of the site location;
determining whether the object is in compliance with logic rules of regulation information based on the object type;
comparing sensor data to the regulation information, wherein the sensor data is received from a sensor communicatively coupled to the computing device, and wherein the sensor data corresponds to an environmental condition of the site location; and
generating a compliance report based on whether the object in the image complies with the regulation information and whether the sensor data complies with the regulation information.
2. The computing device of claim 1 , further comprising prioritizing the object among a plurality of objects that are responsible for severe injuries.
3. The computing device of claim 1 , further comprising identifying the object as a hazardous object based on insurance claims data.
4. The computing device of claim 1 , further comprising combining data about the object with contextual information and semantic information.
5. The computing device of claim 1 , wherein the model of the site location is a building information model (BIM).
6. The computing device of claim 1 , further comprising converting the model to a generic data format.
7. The computing device of claim 1 , wherein the determining whether the object is in compliance with the logic rules of the regulation information further comprises:
converting narratives in the regulation information into the logic rules;
converting the logic rules into rules sets in a form of pseudo codes; and
select a data standard that represents safety requirements in a building information model.
8. The computing device of claim 3 , further comprising, using a text analytics platform to extract unstructured claims data from the insurance claims data.
9. The computing device of claim 1 , wherein the object is included in the image of the site location captured by the camera.
10. The computing device of claim 1 , further comprising receiving, from a database, claims data related to at least one of prior losses or prior claims corresponding to the site location.
11. The computing device of claim 1 , further comprising a display device that displays a prompt to create a bounding box around the object in the image of the site location captured by the camera.
12. The computing device of claim 1 , wherein the compliance report indicates a risk assessment of the site location based on whether the object in the image complies with the regulation information and whether the data sensed by the sensor complies with the regulation information.
13. A method comprising:
capturing, by a camera, an image of an object at a site location;
detecting the object in the image by executing a Convolutional Neural Network (CNN) trained to detect the object in the image using a plurality of convolution layers to extract aspects of the image in a progressively hierarchical fashion by combining features from a previous convolution layer with a next convolution layer through a weighted non-linear function;
identifying, using edge analysis, an object type for the object by determining an object location of the object in the image and matching the object location to a model of the site location;
determining whether the object is in compliance with logic rules of regulation information based on the object type;
comparing sensor data to the regulation information, wherein the sensor data is received from a sensor communicatively coupled to the computing device, and wherein the sensor data corresponds to an environmental condition of the site location; and
generating a compliance report based on whether the object in the image complies with the regulation information and whether the sensor data complies with the regulation information.
14. The method of claim 13 , further comprising prioritizing the object among a plurality of objects that are responsible for severe injuries.
15. The method of claim 13 , further comprising identifying the object as a hazardous object based on insurance claims data.
16. The method of claim 13 , further comprising combining data about the object with contextual information and semantic information.
17. The method of claim 13 , wherein the model of the site location is a building information model (BIM).
18. The method of claim 13 , further comprising receiving, from a database, claims data related to at least one of prior losses or prior claims corresponding to the site location.
19. The method of claim 13 , further comprising displaying a prompt to create a bounding box around the object in the image of the site location captured by the camera.
20. A non-transitory computer-readable medium storing instructions that, when executed by a processor, cause a computing device to perform the following functions:
capturing, by a camera, an image of an object at a site location;
detecting the object in the image by executing a Convolutional Neural Network (CNN) trained to detect the object in the image using a plurality of convolution layers to extract aspects of the image in a progressively hierarchical fashion by combining features from a previous convolution layer with a next convolution layer through a weighted non-linear function;
identifying, using edge analysis, an object type for the object by determining an object location of the object in the image and matching the object location to a model of the site location;
determining whether the object is in compliance with logic rules of regulation information based on the object type;
comparing sensor data to the regulation information, wherein the sensor data is received from a sensor communicatively coupled to the computing device, and wherein the sensor data corresponds to an environmental condition of the site location; and
generating a compliance report based on whether the object in the image complies with the regulation information and whether the sensor data complies with the regulation information.