Enhanced calibration of imaging device output

1. A computer-implemented method comprising:

obtaining, by one or more processors, a current initial image generated by an image generator of an imaging device based on current input signals of sensors included in sensing circuitry of the imaging device; and

applying, by the one or more processors, a transformation model to the current initial image to generate a current transformed image, wherein:

the transformation model is a machine-learning model that has been trained with a set of training data to transform initial images to generate transformed images such that the transformed images more closely resemble reference images generated by a reference image generator of a reference imaging device than the initial images, and

the set of training data includes:

a plurality of initial training images generated by the image generator from a plurality of sets of training input signals, wherein the plurality of sets of training input signals bypass the sensing circuitry of the imaging device and are injected directly to the image generator; and

a plurality of the reference images generated by the reference image generator from the same plurality of sets of training input signals, wherein the plurality of sets of training input signals bypass reference sensing circuitry of the reference imaging device and are injected directly to the reference image generator.

2. The computer-implemented method of claim 1 , further comprising:

generating, by the one or more processors, the set of training data, wherein each initial training image of the plurality of initial training images includes a label identifying an associated reference image of the plurality of the reference images; and

training, by the one or more processors, the transformation model based on the set of training data, using the plurality of initial training images as input and the plurality of the reference images as output.

3. The computer-implemented method of claim 2 , wherein generating the set of training data comprises:

receiving the plurality of sets of training input signals representing possible sensor signals;

injecting the plurality of sets of training input signals to the reference image generator of the reference imaging device;

generating the plurality of the reference images, wherein each reference image of the plurality of the reference images is generated based on one set of the plurality of sets of training input signals; and

storing the plurality of sets of training input signals and the plurality of the reference images in a database in a memory, wherein each set of the plurality of sets of training input signals includes the label identifying the associated reference image of the plurality of the reference images.

4. The computer-implemented method of claim 3 , wherein generating the set of training data further comprises:

receiving the plurality of sets of training input signals from the database in the memory;

injecting the plurality of sets of training input signals to the image generator of the imaging device;

generating the plurality of initial training images, wherein each initial training image of the plurality of initial training images is generated based on one set of training input signals of the plurality of sets of training input signals; and

storing the plurality of initial training images to the database in the memory, wherein each initial training image of the plurality of initial training images includes the label identifying the associated reference image of the plurality of the reference images that was generated from the same set of training input signals of the plurality of sets of training input signals.

5. The computer-implemented method of claim 1 , wherein:

the imaging device comprises a first imaging device of a plurality of imaging devices,

the transformation model comprises a first transformation model of a plurality of transformation models,

each of the plurality of transformation models is implemented by a corresponding imaging device of the plurality of imaging devices, and

the computer-implemented method further comprises, for each respective imaging device of the plurality of imaging devices:

generating, by the one or more processors, a training dataset for the respective imaging device, wherein the training dataset for the respective imaging device includes:

the plurality of initial training images generated by an image generator of the respective imaging device using the plurality of sets of training input signals, and

the plurality of the reference images generated by the reference image generator using the plurality of sets of training input signals; and

training, by the one or more processors, each of the transformation models based on the corresponding training dataset.

6. The computer-implemented method of claim 1 , wherein the machine-learning model of the transformation model comprises one or more convolutional neural networks, each of the one or more convolutional neural networks including one or more convolutional layers, one or more normalization layers, and one or more activation layers.

7. The computer-implemented method of claim 6 , wherein the one or more convolutional neural networks comprises a first convolutional neural network and a second convolutional neural network, the computer-implemented method further comprising:

generating, by the first convolutional neural network, one or more feature maps based on the current initial image; and

generating, by the second convolutional neural network, the current transformed image based on the one or more feature maps.

8. The computer-implemented method of claim 1 , further comprising:

determining, by the one or more processors, whether a difference between the current initial image and the current transformed image exceeds a threshold; and

generating, by the one or more processors, an alert in response to determining that the difference between the current initial image and the current transformed image exceeds the threshold.

9. The computer-implemented method of claim 1 , wherein the transformation model is implemented on a server in communication with the imaging device.

10. The computer-implemented method of claim 1 , wherein the transformation model is trained on one or more cloud computing devices.

11. The computer-implemented method of claim 1 , wherein the sensors one or more of an electrochemical cell, a spectrometer, or an x-ray detector.

12. The computer-implemented method of claim 1 , wherein the current input signals comprise one or more of a single-point output signal or a varying time-series.

13. A computing system comprising:

an imaging device;

at least one memory; and

one or more processors in communication with the at least one memory, the one or more processors configured to:

obtain a current initial image generated by an image generator of the imaging device based on current input signals of sensors included in sensing circuitry of the imaging device; and

apply a transformation model to the current initial image to generate a current transformed image, wherein:

the transformation model is a machine-learning model that has been trained with a set of training data to transform initial images to generate transformed images such that the transformed images more closely resemble reference images generated by a reference image generator of a reference imaging device than the initial images; and

the set of training data includes:

a plurality of initial training images generated by the image generator from a plurality of sets of training input signals, wherein the plurality of sets of training input signals bypass the sensing circuitry of the imaging device and are injected directly to the image generator; and

a plurality of the reference images generated by the reference image generator from the same plurality of sets of training input signals, wherein the plurality of sets of training input signals bypass reference sensing circuitry of the reference imaging device and are injected directly to the reference image generator.

14. The computing system of claim 13 , wherein:

the one or more processors are further configured to:

generate the set of training data, wherein each initial training image of the plurality of initial training images includes a label identifying an associated reference image of the plurality of the reference images; and

train the transformation model based on the set of training data, using the plurality of initial training images as input and the plurality of the reference images as output.

15. The computing system of claim 14 , wherein to generate the set of training data, the one or more processors are configured to:

receive the plurality of sets of training input signals representing possible sensor signals;

inject the plurality of sets of training input signals to the reference image generator of the reference imaging device;

generate the plurality of the reference images, wherein each reference image of the plurality of the reference images is generated based on one set of the plurality of sets of training input signals; and

store the plurality of sets of training input signals and the plurality of the reference images in a database in a memory, wherein each set of the plurality of sets of training input signals includes the label identifying the associated reference image of the plurality of the reference images.

16. The computing system of claim 15 , wherein to generate the set of training data, the one or more processors are further configured to:

receive the plurality of sets of training input signals from the database in the memory;

inject the plurality of sets of training input signals to the image generator of the imaging device;

generate the plurality of initial training images, wherein each initial training image of the plurality of initial training images is generated based on one set of training input signals of the plurality of sets of training input signals; and

store the plurality of initial training images to the database in the memory, wherein each initial training image of the plurality of initial training images includes the label identifying the associated reference image of the plurality of the reference images that was generated from the same set of training input signals of the plurality of sets of training input signals.

17. The computing system of claim 13 , further comprising a plurality of imaging devices, wherein:

the imaging device comprises a first imaging device of the plurality of imaging devices,

the transformation model comprises a first transformation model of a plurality of transformation models,

each of the plurality of transformation models is implemented by a corresponding imaging device of the plurality of imaging devices, and

for each respective imaging device of the plurality of imaging devices, the one or more processors are further configured to:

generate a training dataset for the respective imaging device, wherein the training dataset for the respective imaging device includes:

the plurality of initial training images generated by an image generator of the respective imaging device using the plurality of sets of training input signals, and

the plurality of the reference images generated by the reference image generator using the plurality of sets of training input signals; and

train each of the transformation models based on the corresponding training dataset.

18. The computing system of claim 13 , wherein:

the machine learning model of the transformation model comprises at least a first convolutional neural network and a second convolutional neural network,

each convolutional neural network of the machine-learning model includes one or more convolutional layers, one or more normalization layers, and one or more activation layers; and

the one or more processors are further configured to:

generate, by the first convolutional neural network, one or more feature maps based on the current initial image; and

generate, by the second convolutional neural network, the current transformed image based on the one or more feature maps.

19. The computing system of claim 13 , wherein the one or more processors are further configured to:

determine whether a difference between the current initial image and the current transformed image exceeds a threshold; and

generate an alert in response to determining that the difference between the current initial image and the current transformed image exceeds the threshold.

20. A computer-implemented method comprising:

generating, by one or more processors, a plurality of sets of training input signals representing possible sensor signals from sensors of a reference imaging device;

generating, by the one or more processors, a plurality of reference images by injecting the plurality of sets of training input signals to a reference image generator of the reference imaging device;

generating, by the one or more processors, a plurality of initial training images by injecting the plurality of sets of training input signals to an image generator of an imaging device;

storing, by the one or more processors, the plurality of initial training images to a database in a memory, wherein:

each initial training image of the plurality of initial training images includes a label identifying an associated reference image of the plurality of reference images that was generated from the same set of training input signals of the plurality of sets of training input signals; and

a set of training data includes the plurality of initial training images and the plurality of reference images; and

training, by the one or more processors, a machine-learning model based on the set of training data to enable the machine-learning model to transform initial images to generate transformed images such that the transformed images more closely resemble the plurality of reference images.