Claims:We claim:
1. A method of implementing neural network models on edge devices in an Internet of Things (IoT) network, the method comprising:
receiving, by an optimizing device, a neural network model trained and configured to detect objects from images, wherein the neural network model is a Convolutional Neural Network (CNN);
iteratively assigning, by the optimizing device, a new value to each of a plurality of parameters associated with the neural network model to generate a re-configured neural network model in each iteration;
deploying for a current iteration, by the optimizing device, the re-configured neural network on an edge device to identify at least one object in an image;
computing, by the optimizing device, for the current iteration, a trade-off value based on a detection accuracy associated with the at least one object detected in the image and resource utilization data associated with the edge device; and
selecting, by the optimizing device, the re-configured neural network model of the current iteration, based on the trade-off value calculated for the current iteration.

2. The method of claim 1, wherein the selecting further comprises:
comparing the trade-off value calculated for the current iteration with a predefined threshold; and
selecting the re-configured neural network model of the current iteration, when the trade-off value is greater than or equal to the predefined threshold.

3. The method of claim 1 further comprising:
rendering a parameter interface to a user, wherein the parameter interface is configured to receive values assigned to one or more of the plurality of parameters from the user, wherein the values assigned to the one or more of the plurality of parameters is used for iteratively assigning the new value to each of a plurality of parameters associated with the neural network model.

4. The method of claim 1, wherein the resource usage data associated with the edge device corresponds to one or more of a loss function, memory usage, and Millions of Instructions Per Second (MIPS) usage of the re-configured neural network model for the edge device.

5. The method of claim 1, wherein the plurality of parameters comprises at least one of batch_normalize, a filter, a size, a stride, a pad, an activation, a maxpool_size, a maxpool_stride, a bit-width, a memory allocation, a batchsize, an input image resolution, an input image height, an input image width, a channel, a learning rate, an epoch, a decay of learning rate, a quantization of weights and biases, a gridsize, a bounding_box_ per_grid_cell, an output_class, and a weight for Loss function.

6. An optimizing device for implementing neural network models on edge devices in an Internet of Things (IoT) network, the optimizing device comprising:
a processor; and
a memory communicatively coupled to the processor, wherein the memory stores processor instructions, which, on execution, causes the processor to:
receive a neural network model trained and configured to detect objects from images, wherein the neural network model is a Convolutional Neural Network (CNN);
iteratively assign a new value to each of a plurality of parameters associated with the neural network model to generate a re-configured neural network model in each iteration;
deploy, for a current iteration, the re-configured neural network on an edge device to identify at least one object in an image;
compute for the current iteration, a trade-off value based on a detection accuracy associated with the at least one object detected in the image and resource utilization data associated with the edge device; and
select the re-configured neural network model of the current iteration, based on the trade-off value calculated for the current iteration.

7. The optimizing device of claim 6, wherein the processor instructions further cause the processor to:
receive an image from an image capturing device within the edge device; and
detect one or more objects in the received image, using the re-configured neural network model of the current iteration.

8. The optimizing device of claim 6, wherein the selecting further comprises:
comparing the trade-off value calculated for the current iteration with a predefined threshold; and
selecting the re-configured neural network model of the current iteration, when the trade-off value is greater than or equal to the predefined threshold.

9. The optimizing device of claim 6, wherein the processor instructions further cause the processor to render a parameter interface to a user, wherein the parameter interface is configured to receive values assigned to one or more of the plurality of parameters from the user.

10. The optimizing device of claim 9, wherein the values assigned to the one or more of the plurality of parameters is used for iteratively assigning the new value to each of a plurality of parameters associated with the neural network model.

Dated this 3rd day of June, 2019

R Ramya Rao
Of K&S Partners
Agent for the Applicant
IN/PA-1607
, Description:Technical Field
[001] This disclosure relates generally to implementing neural network models, and more particularly to a method and system of implementing neural network models on edge devices in Internet of Things (IoT) networks.