Systems, methods, and computer readable media for predictive analytics and change detection from remotely sensed imagery
Systems and methods are provided for automatically detecting a change in a feature. For example, a system includes a memory and a processor configured to analyze a change associated with a feature over a period of time using a plurality of remotely sensed time series images. Upon execution, the system would receive a plurality of remotely sensed time series images, extract a feature from the plurality of remotely sensed time series images, generate at least two time series feature vectors based on the feature, where the at least two time series feature vectors correspond to the feature at two different times, create a neural network model configured to predict a change in the feature at a specified time, and determine, using the neural network model, the change in the feature at a specified time based on a change between the at least two time series feature vectors.
1. A method of detecting a change in a feature in remotely sensed time series images, the method comprising:
receiving, at a server, a plurality of remotely sensed time series images;
extracting, at the server, a feature from the plurality of remotely sensed time series images;
generating, at the server, at least two time series feature vectors based on the feature, wherein the at least two time series feature vectors correspond to the feature at two different times;
creating, at the server, a neural network model configured to predict a change in the feature at a specified time; and
determining, at the server using the neural network model, the change in the feature at the specified time based on a change between the at least two time series feature vectors, wherein the neural network model receives as input the at least two time series feature vectors.
2. The method of claim 1 , wherein the feature is extracted by a convolutional neural network, and wherein the at least two time series feature vectors are generated by the convolutional neural network.
3. The method of claim 1 , wherein the neural network model is based on a neural network that comprises at least one of a recurrent neural network or a convolutional neural network.
4. The method of claim 1 , wherein the plurality of remotely sensed time series images are captured at regular intervals.
5. The method of claim 1 , wherein the plurality of remotely sensed time series images are captured at non-regular intervals.
6. The method of claim 1 , wherein the feature is at least one of a property condition, a neighborhood condition, a built environment, a vegetation state, a topology, a roof, a building, a tree, or a body of water.
7. The method of claim 1 , wherein the feature is defined by an aggregated state based on one or more sub-states of one or more partial features.
8. The method of claim 1 further comprising determining, at the server, a plurality of time series scores, each of which corresponds to the feature.
9. The method of claim 8 , wherein the change in the feature is determined based on the plurality of time series scores.
10. The method of claim 8 , wherein an insignificant change in the feature is discarded.
11. A server for detecting a change in a feature in remotely sensed time series images, the server comprising:
a memory that stores a module; and
a processor configured to run the module stored in the memory that is configured to cause the processor to:
receive a plurality of remotely sensed time series images;
extract a feature from the plurality of remotely sensed time series images;
generate at least two time series feature vectors based on the feature, wherein the at least two time series feature vectors correspond to the feature at two different times;
create a neural network model configured to predict a change in the feature at a specified time; and
determine, using the neural network model, the change in the feature at the specified time based on a change between the at least two time series feature vectors, wherein the neural network model receives as input the at least two time series feature vectors.
12. The server of claim 11 , wherein the feature is extracted by a convolutional neural network, and wherein the at least two time series feature vectors are generated by the convolutional neural network.
13. The server of claim 11 , wherein the neural network model is based on a neural network that comprises at least one of a recurrent neural network or a convolutional neural network.
14. The server of claim 11 , wherein the plurality of remotely sensed time series images is captured at regular intervals.
15. The server of claim 11 , wherein the plurality of remotely sensed time series images is captured at non-regular intervals.
16. The server of claim 11 , wherein the feature is at least one of a property condition, a neighborhood condition, a built environment, a vegetation state, a topology, a roof, a building, a tree, or a body of water.
17. The server of claim 11 , wherein the feature is defined by an aggregated state based on one or more sub-states one or more partial features.
18. The server of claim 11 , wherein the module stored in the memory is further configured to cause the processor to determine a plurality of time series scores, each of which corresponds to the feature.
19. The server of claim 18 , wherein the change in the feature is determined based on the plurality of time series scores.
20. The server of claim 18 , wherein an insignificant change is discarded.
21. A non-transitory computer readable medium storing executable instructions operable for detecting a change in a feature in remotely sensed time series images to cause a processor to perform operations comprising:
receiving a plurality of remotely sensed time series images;
extracting a feature from the plurality of remotely sensed time series images;
generating at least two time series feature vectors based on the feature, wherein the at least two time series feature vectors correspond to the feature at two different times;
creating a neural network model configured to predict a change in the feature at a specified time; and
determining, using the neural network model, the change in the feature at a specified time based on a change between the at least two time series feature vectors, wherein the neural network model receives as input the at least two time series feature vectors.