SYSTEMS AND METHODS OF CAMERA-BASED MACHINE VISION MODEL RESOURCE ALLOCATION

1 . A camera, comprising:

an image capture device to output one or more images; and

one or more processors coupled with memory, the one or more processors to:

detect a feature represented in a first image of the one or more images;

select, based on the feature, a machine vision model to process at least one of the first image or a second image of the one or more images;

allocate, to the machine vision model, based on a characteristic of the machine vision model, a portion of the memory; and

execute the machine vision model using the portion of the memory to process the at least one of the first image or the second image.

2 . The camera of claim 1 , wherein the machine vision model is a second machine vision model, the portion of the memory is a second portion of the memory, and the one or more processors are to:

detect the feature represented in the first image by causing a first machine vision model to execute on a first portion of memory.

3 . The camera of claim 1 , wherein the one or more processors are to determine an amount of the portion of memory to allocate to the machine vision model based on at least one of a frame rate associated with the one or more images or a resolution of the one or more images.

4 . The camera of claim 1 , wherein the one or more processors are to:

monitor a utilization of the memory by the machine vision model; and

output an alert responsive to the utilization exceeding a threshold.

5 . The camera of claim 1 , wherein the one or more processors are to:

monitor a utilization of the memory by the machine vision model; and

modify operation of the machine vision model responsive to the utilization exceeding a threshold.

6 . The camera of claim 1 , wherein the portion of the memory is a second portion of the memory, and one or more processors are to:

retrieve a data structure corresponding to the machine vision model from a first portion of the memory; and

deploy the machine vision model on the second portion of the memory using the retrieved data structure.

7 . The camera of claim 1 , wherein the one or more processors are to select the machine vision model from a plurality of machine vision models.

8 . The camera of claim 1 , wherein the machine vision model comprises a neural network to perform at least one of object detection, object tracking, or alarm generation.

9 . The camera of claim 1 , wherein the one or more processors are to allocate the portion of memory to include at least one of a minimum threshold amount of the memory or a maximum threshold amount of the memory.

10 . The camera of claim 1 , wherein the image capture device is to output the one or more images according to a predetermined quantization, and the machine vision model is configured to process the one or more images according to the predetermined quantization.

11 . The camera of claim 1 , wherein the image capture device is to output the first image and the second image as a sequence of images at a frame rate.

12 . A method, comprising:

selecting, by processing circuitry of a camera, a machine vision model to process one or more images from an image capture device of the camera;

allocating a portion of memory of the camera, by the processing circuitry, based on a characteristic of at least one of the one or more images or the machine vision model; and

executing, by the processing circuitry, the machine vision model on the portion of the memory to process the one or more images.

13 . The method of claim 12 , wherein the characteristic comprises at least one of a frame rate associated with the one or more images or a resolution of the one or more images.

14 . The method of claim 12 , wherein the machine vision model is a second machine vision model, the portion of the memory is a second portion of the memory, and the method further comprises:

detecting, by the processing circuitry, the feature represented in the first image by causing a first machine vision model to execute on a first portion of memory.

15 . The method of claim 12 , further comprising:

monitoring, by the processing circuitry, a utilization of the portion of the memory by the machine vision model; and

output an alert responsive to the utilization exceeding a threshold.

16 . The method of claim 12 , wherein the portion of the memory is a second portion, the method further comprising:

retrieving, by the processing circuitry, a plurality of weights and biases corresponding to the machine vision model from a first portion of the memory; and

deploying, by the processing circuitry, the machine vision model on the second portion of the memory using the plurality of weights and biases.

17 . The method of claim 12 , wherein the machine vision model comprises a second machine vision model, the method further comprising:

detecting, by the processing circuitry, using a first machine vision model, a feature of an object represented in the one or more images;

selecting, by the processing circuitry, the second machine vision model based on the feature; and

processing, by the second machine vision model, the feature to detect a characteristic of the feature.

18 . A system, comprising:

one or more processors to:

detect an object in image data from an image capture device, the object having a class;

select, based on the class, a machine vision model to process the image stream;

determine a target memory usage for the machine vision model; and

execute the machine vision model, on a portion of memory corresponding to the target memory usage, to generate an output regarding the object based on the image data.

19 . The system of claim 18 , wherein the one or more processors are to:

monitor utilization of the portion of memory by the machine vision model; and

update operation of the machine vision model responsive to the utilization exceeding a threshold.

20 . The system of claim 18 , wherein the machine vision model is to determine a state of the object based on the image data.