Systems and methods for distributed data analytics
The invention provides systems and method for generating device-specific artificial neural network (ANN) models for distribution across user devices. Sample datasets are collected from devices in a particular environment or use case and include predictions by device-specific ANN models executing the user devices. The received datasets are used with existing datasets and stored ANN models to generate updated device-specific ANN models from each of the stored instances of the device ANN models based on the training data.
1. A method for identifying an object of interest in an image file, the method comprising:
receiving one or more image files, each image file of the one or more images files including an object of interest;
applying a non-binary ground truth label to each image file, the non-binary ground truth label comprising a distribution of pixel-specific salience values indicating the likelihood that a particular pixel is part of the object of interest;
generating a three-dimensional salience surface map having an x axis, y axis and z axis, wherein the x axis and the y axis values define a location of the pixels in the image, and the z axis value being the pixel-specific salience values;
selecting a curve shape from a library of curve shapes;
applying the curve shape to the salience surface map to determine a fit between the curve shape and the three-dimensional surface; and
determining whether the image file includes the object of interest based on the fit.
2. The method of claim 1 , wherein the curve shape is selected based on the object of interest.
3. The method of claim 1 , wherein the curve shape is selected from one of a Gaussian distribution, Poisson distribution, and hybrid distribution.
4. The method of claim 1 , further comprising adding the image file to a library of imagefiles for use in training an artificial neural network (ANN).
5. The method of claim 4 , wherein the ANN is trained to identify the object of interest in subsequent media files.
6. The method of claim 4 , wherein the ANN is trained to segment objects within in subsequent media files.
7. A system for identifying an object of interest in an image file, the system comprising:
one or more processors; and
a memory coupled with the one or more processors wherein the one or more processors executes computer-executable instructions stored in the memory, that when executed:
receives one or more image files, each image file including an object of interest;
applies a non-binary ground truth label to each image file, the non-binary ground truth label comprising a distribution of pixel-specific salience values indicating the likelihood that a particular pixel is part of the object of interest;
generates a three-dimensional salience surface map having an x axis, y axis and z axis, wherein the x axis and the y axis values define a location of the pixels in the image, and the z axis value being the pixel-specific salience values;
selects a curve shape from a library of curve shapes and applying the curve shape to the salience surface map to determine a fit between the curve shape and the three-dimensional surface; and
determines whether the image file includes the object of interest based on the fit.
8. The system of claim 7 , wherein the curve shape is selected based on the object of interest.
9. The system of claim 7 , wherein the curve shape is selected from one of a Gaussian distribution, a Poisson distribution, and a hybrid distribution.
10. The system of claim 7 , wherein execution of the instructions further adds the image file to a library of image files for use in training an artificial neural network (ANN).
11. The system of claim 10 , wherein the ANN is trained to identify the object of interest in subsequent media files.
12. The system of claim 10 , wherein the ANN is trained to segment objects within in subsequent media files.
13. A non-transitory computer-readable storage medium storing computer-readable instructions that, when executed by a computer, cause the computer to perform a method, the method comprising:
receiving one or more image files, each image file of the one or more images files including an object of interest;
applying a non-binary ground truth label to each image file, the non-binary ground truth label comprising a distribution of pixel-specific salience values indicating the likelihood that a particular pixel is part of the object of interest;
generating a three-dimensional salience surface map having an x axis, y axis and z axis, wherein the x axis and the y axis values define a location of the pixels in the image, and the z axis value being the pixel-specific salience values;
selecting a curve shape from a library of curve shapes;
applying the curve shape to the salience surface map to determine a fit between the curve shape and the three-dimensional surface; and
determining whether the image file includes the object of interest based on the fit.