Estimating DeltaV from images
AssigneeCcc Intelligent Solutions Inc.
InventorsZia Ghiasi, Taha Khademinejad, Surendran Narayanamoorthy, Mohamad Reza Rooholamini
Life of this patent
Estimating the term…
Abstract
An artificial intelligence (AI) deep learning model in an estimator may be trained by inputting the photos of known accidents to create a trained model. The trained model in the estimator may take First Notice of Loss (FNOL) data to predict the value of DeltaV and PDoF for a plurality of objects in a new accident.
Claims19
A method of determining a plurality of objects' change in velocity from one or more images of the objects comprising:
- receiving at least one or more images of the objects;
- submitting the at least one or more images of the objects to an estimator;
- using the estimator to:
- estimate change of each of the object's velocity indicated by the at the least one or more images wherein the estimator:
- accepts one or more training images of a training object and related change in velocity of the training object from impact data;
- and
- submits the one or more training images and related change in velocity of the training object from impact data to a computer based learning algorithm to train the estimator to predict an object's change in velocity from one or more images of the object wherein the estimated change of velocity comprises a DeltaV estimate wherein Delta V indicates a change in the object's velocity and a principle direction of force (PDoF).
A non-transitory computer readable medium comprising computer executable instructions that physically configure a processor, the computer executable instruction comprising instructions for determining a DeltaV for objects from one or more images of the objects wherein DeltaV indicates a change in the object's velocity and a principle direction of force (PDoF) comprising:
- receiving at least one or more images of the objects;
- determining features of the at least one or more images;
- submitting the at least one or more images of the objects and features determined to an estimator;
- using the estimator to:
- estimate change of each of the object's velocity indicated by the one or more images wherein the estimator:
- accepts one or more training images of a training object and related change in velocity of the training object from impact data;
- determines features for each of the one or more training images and related change in velocity of the training object from impact data and
- submits the one or more training images, related change in velocity of the training object from impact data and the features of the one or more training images to a computer based learning algorithm to train the estimator to predict an object's DeltaV from one or more images of the objects and the features determined for the at least one or more images.
A computer system comprising:
- a processor that is physically configured according to computer executable instructions,
- a memory in communication with the processor;
- and
- an input-output circuit in communication with the processor,
- the computer executable instruction comprising instructions for determining a DeltaV for a plurality of objects from one or more images of the object wherein DeltaV indicates a change in each of the object's velocity and a principle direction of force (PDoF) comprising:
- receiving at least one or more images of the object;
- determines features of the at least one or more images;
- submitting the at least one or more images of the object and features determined to an estimator;
- using the estimator to:
- estimate change of the objects velocity indicated by the at the least one or more images wherein the estimator:
- accepts one or more training images of a training object and related change in velocity of the training object from impact data;
- determines features for each of the one or more training images and related change in velocity of the training object from impact data and submits the one or more training images, related change in velocity of the training object from impact data and the features of the one or more training images to a computer based learning algorithm to train the estimator to predict an object's DeltaV from one or more images of the object and the features determined for the at least one or more images.
Dependencies are inferred from the claim text.