Systems and methods for estimating object weight using camera images
In some aspects, the techniques described herein relate to a method for object weight estimation, including: training an object identification model; receiving one or more images from a camera positioned on a machine engaged in a lifting operation; feeding each of the one or more images into the object identification model; receiving, from the trained object identification model, a predicted object corresponding to the object to be lifted by the machine and a probability that the predicted object corresponds to the object being lifted by the machine; and estimating a weight of the object based on at least the predicted object and comparing the estimated weight to lifting parameters associated with the machine when the probability is greater than a selected confidence threshold.
1. A method for object weight estimation, comprising:
training an object identification model;
receiving one or more images from a camera positioned on a machine engaged in a lifting operation;
feeding each of the one or more images into the object identification model;
receiving, from the trained object identification model, a predicted object corresponding to the object to be lifted by the machine and a probability that the predicted object corresponds to the object being lifted by the machine; and
estimating a weight of the object based on at least the predicted object and comparing the estimated weight to lifting parameters associated with the machine when the probability is greater than a selected confidence threshold.
2. The method of claim 1 , further comprising providing a recommendation when the estimated weight exceeds one or more of the lifting parameters.
3. The method of claim 1 , wherein training the object identification model, comprises:
collecting a plurality of training images for each of a plurality of objects;
labeling each of the plurality of training images with a corresponding one of a plurality of object identifiers; and
training an object neural network with the plurality of labeled training images.
4. The method of claim 1 , further comprising, when the probability is less than the selected confidence threshold, receiving additional images from the camera and feeding the additional images into the trained object identification model until the probability is greater than the selected confidence threshold.
5. The method of claim 1 , wherein the selected confidence threshold is selected based on a type of the predicted object.
6. The method of claim 1 , wherein the weight of the object is estimated based on a color of the object.
7. The method of claim 1 , further comprising indicating when the estimated weight exceeds one or more of the lifting parameters.
8. A system for object weight estimation, comprising:
one or more cameras positioned on a machine;
one or more processors; and
one or more memory devices having stored thereon instructions that when executed by the one or more processors cause the one or more processors to:
train an object identification model;
receive one or more images from the one or more cameras;
feed each of the one or more images into the object identification model;
receive, from the trained object identification model, a predicted object corresponding to an object to be lifted by the machine and a probability that the predicted object corresponds to the object to be lifted by the machine; and
estimate a weight of the object based on at least the predicted object and compare the estimated weight to lifting parameters associated with the machine when the probability is greater than a selected confidence threshold.
9. The system of claim 8 , further comprising providing a recommendation when the estimated weight exceeds one or more of the lifting parameters.
10. The system of claim 8 , wherein training the object identification model, comprises:
collecting a plurality of training images for each of a plurality of objects;
labeling each of the plurality of training images with a corresponding one of a plurality of object identifiers; and
training an object neural network with the plurality of labeled training images.
11. The system of claim 8 , further comprising, when the probability is less than the selected confidence threshold, receiving additional images from the camera and feeding the additional images into the trained object identification model until the probability is greater than the selected confidence threshold.
12. The system of claim 8 , wherein the selected confidence threshold is selected based on a type of the predicted object.
13. The system of claim 8 , wherein the weight of the object is estimated based on a color of the object.
14. The system of claim 8 , further comprising an indicator on the machine communicatively coupled to the one or more processors and operative to indicate when the estimated weight exceeds one or more of the lifting parameters.
15. One or more non-transitory computer-readable media storing computer-executable instructions that, when executed by one or more processors, cause the one or more processors to perform operations comprising:
training an object identification model;
receiving one or more images from a camera positioned on a machine engaged in a lifting operation;
feeding each of the one or more images into the object identification model;
receiving, from the trained object identification model, a predicted object corresponding to the object to be lifted by the machine and a probability that the predicted object corresponds to the object being lifted by the machine; and
estimating a weight of the object based on at least the predicted object and comparing the estimated weight to lifting parameters associated with the machine when the probability is greater than a selected confidence threshold.
16. The one or more non-transitory computer-readable media of claim 15 , further comprising providing a recommendation when the estimated weight exceeds one or more of the lifting parameters.
17. The one or more non-transitory computer-readable media of claim 15 , wherein training the object identification model, comprises:
collecting a plurality of training images for each of a plurality of objects;
labeling each of the plurality of training images with a corresponding one of a plurality of object identifiers; and
training an object neural network with the plurality of labeled training images.
18. The one or more non-transitory computer-readable media of claim 15 , further comprising, when the probability is less than the selected confidence threshold, receiving additional images from the camera and feeding the additional images into the trained object identification model until the probability is greater than the selected confidence threshold.
19. The one or more non-transitory computer-readable media of claim 15 , wherein the selected confidence threshold is selected based on a type of the predicted object.
20. The one or more non-transitory computer-readable media of claim 15 , wherein the weight of the object is estimated based on a color of the object.