System and method for facilitating training of a prediction model to estimate a user vehicle damage tolerance

Abstract

Some embodiments relate to techniques for facilitating training of a prediction model for estimating a threshold score for a user. In some embodiments, a first image of at least a first portion of a first vehicle may be provided to a client device, where the first image may be associated with a first damage score. From the client device, a user-provided score for the first image may be received. Based on the user-provided score, a second image of at least a second portion of a second vehicle may be provided to the client device, where the second image may be associated with a second damage score. Training data may be generated based on the first damage score and the second damage score, and the training data may be provided to a prediction model to train the prediction model to estimate a threshold score for a user.

What is claimed is: 
     
         1 . A non-transitory computer-readable medium storing computer program instructions that, when executed by one or more processors, effectuate operations comprising:
 selecting a first image depicting at least a portion of a first vehicle based on a score for a second image depicting at least a portion of a second vehicle, wherein the first image is associated with a first pre-computed score, and the second image is associated with a second pre-computed score;   providing the first image to a client device;   receiving, from the client device, an indication of whether to purchase a vehicle having a same or similar amount of damage as the first vehicle;   responsive to determining that the first pre-computed score and a score for the first image satisfy a threshold similarity condition, wherein the score for the first image is associated with the indication, providing the first pre-computed score or the score for the first image to a first prediction model to estimate a threshold amount of vehicle damage a user tolerates when determining whether to purchase a vehicle, wherein the threshold similarity condition comprises the first pre-computed score and the score for the first image differing by less than a threshold value; and   responsive to determining that the first pre-computed score and the score for the first image fail to satisfy the threshold similarity condition, selecting a third image depicting at least a portion of a third vehicle based on the score for the first image.   
     
     
         2 . The non-transitory computer-readable medium of  claim 1 , wherein the operations further comprise:
 selecting, in response to a request for content regarding vehicles available for purchase, a plurality of images depicting vehicles based on the threshold amount of vehicle damage; and   providing at least some of the plurality of images to one or more users associated with the request.   
     
     
         3 . The non-transitory computer-readable medium of  claim 1 , wherein the operations further comprise:
 generating training data based on a plurality of images, wherein a first subset of images of the plurality of images each depict one or more portions of a corresponding vehicle including damage, and wherein a second subset of images of the plurality of images each depict one or more portions of a corresponding vehicle not including damage; and   providing the training data to a second prediction model to be trained to estimate a pre-computed score for a given image depicting at least one portion of a vehicle including damage, wherein the first pre-computed score for the first image and the second pre-computed score for the second image are determined with the second prediction model.   
     
     
         4 . The non-transitory computer-readable medium of  claim 1 , wherein:
 the score for the second image is received from the client device in response to second image being provided with the client device; and   the first image is provided to the client device subsequent to second image being provided to the client device and responsive to determining that the second pre-computed score and the score for the second image fail to satisfy the threshold similarity condition.   
     
     
         5 . A non-transitory computer-readable medium storing computer program instructions that, when executed by one or more processors, effectuate operations comprising:
 selecting a first image depicting at least a portion of a first vehicle based on a score for a second image depicting at least a portion of a second vehicle, wherein the first image is associated with a first pre-computed score, and the second image is associated with a second pre-computed score;   providing the first image to a client device;   receiving, from the client device, an indication related to purchasing a vehicle having a same or similar amount of damage as the first vehicle; and   responsive to determining, based on the first pre-computed score and the indication, that a threshold condition is satisfied, providing the first pre-computed score or a score for the first image to a prediction model, wherein the score for the first image is determined based on the indication.   
     
     
         6 . The non-transitory computer-readable medium of  claim 5 , wherein the operations further comprise:
 responsive to determining that the first pre-computed score and the indication fail to satisfy the threshold condition, selecting a third image depicting at least a portion of a third vehicle based on the indication.   
     
     
         7 . The non-transitory computer-readable medium of  claim 5 , wherein the prediction model is trained to estimate a threshold score for a user indicating a vehicle damage tolerance of the user when determining whether to purchase a vehicle. 
     
     
         8 . The non-transitory computer-readable medium of  claim 7 , wherein the operations further comprise:
 providing, based on the threshold score, a plurality of images depicting vehicles to at least one user that submitted a request for vehicles available for purchase.   
     
     
         9 . The non-transitory computer-readable medium of  claim 5 , wherein the indication related to purchasing the vehicle comprises the score for the first image, approval of the first pre-computed score for the first vehicle, or disapproval of the first pre-computed score for the first vehicle. 
     
     
         10 . The non-transitory computer-readable medium of  claim 5 , wherein the operations further comprise:
 generating training data based on a plurality of images depicting vehicles including damage and vehicles not including damage; and   providing the training data to an additional prediction model to determine a pre-computed score for a given image, wherein the first pre-computed score and the second pre-computed score are determined with the additional prediction model.   
     
     
         11 . The non-transitory computer-readable medium of  claim 5 , wherein the operations further comprise:
 responsive to determining, based on the first pre-computed score and the indication, that the threshold condition is not satisfied:
 (a) selecting a third image depicting at least a portion of a third vehicle based on the score for the first image; 
 (b) providing the third image to the client device; and 
 (c) receiving, from the client device, an additional indication related to purchasing a vehicle having a same or similar amount of damage as the third vehicle, wherein steps (a)-(c) are iteratively repeated until a selected image and a score associated with the selected image satisfy the threshold condition, wherein the score is obtained in response to providing the selected image to the client device. 
   
     
     
         12 . The non-transitory computer-readable medium of  claim 5 , wherein the threshold condition comprises the first pre-computed score and the score for the first image differing by less than a threshold value. 
     
     
         13 . A method implemented by one or more processors executing one or more computer program instructions that, when executed, perform the method, the method comprising:
 selecting a first image depicting at least a portion of a first vehicle based on a score for a second image depicting at least a portion of a second vehicle, wherein the first image is associated with a first pre-computed score, and the second image is associated with a second pre-computed score;   providing the first image to a client device;   receiving, from the client device, an indication related to purchasing a vehicle having a same or similar amount of damage as the first vehicle; and   responsive to determining, based on the first pre-computed score and the indication, that a threshold condition is satisfied, providing the first pre-computed score or a score for the first image to a prediction model, wherein the score for the first image is determined based on the indication.   
     
     
         14 . The method of  claim 13 , further comprising:
 responsive to determining that the first pre-computed score and the indication fail to satisfy the threshold condition, selecting a third image depicting at least a portion of a third vehicle based on the indication.   
     
     
         15 . The method of  claim 13 , wherein the prediction model is trained to estimate a threshold score for a user indicating a vehicle damage tolerance of the user when determining whether to purchase a vehicle. 
     
     
         16 . The method of  claim 15 , further comprising:
 providing, based on the threshold score, a plurality of images depicting vehicles to at least one user that submitted a request for vehicles available for purchase.   
     
     
         17 . The method of  claim 13 , wherein the indication related to purchasing the vehicle comprises the score for the first image, approval of the first pre-computed score for the first vehicle, or disapproval of the first pre-computed score for the first vehicle. 
     
     
         18 . The method of  claim 13 , further comprising:
 generating training data based on a plurality of images depicting vehicles including damage and vehicles not including damage; and   providing the training data to another prediction model to determine a pre-computed score for a given image, wherein the first pre-computed score and the second pre-computed score are determined with the other prediction model.   
     
     
         19 . The method of  claim 13 , further comprising:
 responsive to determining, based on the first pre-computed score and the indication, that the threshold condition is not satisfied:
 (a) selecting a third image depicting at least a portion of a third vehicle based on the score for the first image; 
 (b) providing the third image to the client device; and 
 (c) receiving, from the client device, an additional indication related to purchasing a vehicle having a same or similar amount of damage as the third vehicle, wherein steps (a)-(c) are iteratively repeated until a selected image and a score associated with the selected image satisfy the threshold condition, wherein the score is obtained in response to providing the selected image to the client device. 
   
     
     
         20 . The method of  claim 13 , wherein the threshold condition comprises the first pre-computed score and the score for the first image differing by less than a threshold value.