Detailed damage determination with image cropping
The present invention relates to the determination of damage to portions of a vehicle. More particularly, the present invention relates to determining whether each part of a vehicle should be classified as damaged or undamaged and optionally the severity of the damage to each part of the damaged vehicle including preserving the quality of the input images of the damage to the vehicle. Aspects and/or embodiments seek to provide a computer-implemented method for determining damage states of each part of a damaged vehicle, indicating whether each part of the vehicle is damaged or undamaged and optionally the severity of the damage to each part of the damaged vehicle, using images of the damage to the vehicle and trained models to assess the damage indicated in the images of the damaged vehicle, including preserving the quality and/or resolution of the images of the damaged vehicle.
1. A computer-implemented method for determining one or more damage states to one or more parts of a vehicle, comprising:
receiving one or more images of the vehicle;
generating a plurality of randomized cropped portions of the one or more images of the vehicle;
determining, per part, one or more classifications for each of the plurality of randomized cropped portions using one or more trained models, wherein each classification comprises at least one indication of damage to at least one part;
determining one or more damage states of each of the parts of the vehicle using the one or more classifications for each of the plurality of randomized cropped portions; and
outputting the one or more damage states of each of the parts of the vehicle.
2. The computer-implemented method of claim 1 , further comprising pre-processing the received one or more images of the vehicle to determine which of the one or more images of the vehicle meet a predetermined crop suitability threshold.
3. The computer-implemented method of claim 2 , wherein the predetermined crop suitability threshold comprises determining whether the content of each of the one or more images comprises one or more close ups of the vehicle.
4. The computer-implemented method of claim 2 , wherein the predetermined crop suitability threshold comprises determining whether the content of each of the one or more images comprises one or more damage areas.
5. The computer-implemented method of claim 2 , wherein pre-processing the received one or more images of the vehicle comprises one or more trained preliminary classification models, optionally the one or more trained models comprise convolutional neural networks.
6. The computer-implemented method of claim 1 , further comprising determining a plurality of parts of the damaged vehicle that are represented in the plurality of images of the vehicle comprises the use of a plurality of classifiers.
7. The computer-implemented method of claim 6 , wherein each one of the plurality of classifiers is operable to detect each of the parts of the vehicle.
8. The computer-implemented method of claim 1 , wherein the damage states of parts of the vehicle are determined as one or more quantitative values.
9. The computer-implemented method of claim 1 , wherein the image resolution of each of the received plurality of images is maintained.
10. The computer-implemented method of claim 1 , wherein the one or more trained models comprise any one or any combination of a neural network; a convolutional neural network; or a recurrent neural network.
11. The computer-implemented method of claim 1 , wherein determining the one or more classifications or determining the one or more damage states comprises using a multi-instance learned approach.
12. The computer-implemented method of claim 1 , further comprising segmenting each of the plurality of images to produce one or more segmentations and wherein determining the one or more classifications uses the one or more segmentations determined for each of the plurality of images.