System, method, and non-transitory computer readable storage medium for image recognition based on convolutional neural networks

1. A recognition system, comprising:

an image capturing device configured to generate a Mth-layer calculation result based on image data and a convolutional neural network (CNN), and configured to transmit feature information associated with the Mth-layer calculation result, wherein M is a positive integer, M is equal to or greater than 1, M is less than or equal to N, and N is a predetermined positive integer; and

a server coupled to the image capturing device and configured to receive the feature information, wherein the server generates a Kth-layer calculation result based on the feature information and the CNN by an iterative method when M is less than N, wherein K is a positive integer which is greater than M and less than or equal to N, and the server generates a first recognition result associated with the image data based on the Kth-layer calculation result and a first recognition model when K is equal to N, in order to recognize the image data.

2. The recognition system of claim 1 , wherein when M is equal to N, the Mth-layer calculation result is configured to be as the feature information and the server generates the first recognition result based on the feature information and the first recognition model.

3. The recognition system of claim 1 , wherein the server is further configured to generate a second recognition result associated with the image data based on the first recognition result and a second recognition model, and the second recognition model is different from the first recognition model.

4. The recognition system of claim 1 , wherein the Mth-layer calculation result comprises a convolution result, a pooling result, an activation result, or a deconvolution result.

5. The recognition system of claim 1 , wherein the image capturing device is further configured to determine whether a threshold time is reached or not, the image capturing device transmits the Mth-layer calculation result to the server to be as the feature information when M is less than N and the threshold time is reached, and the image capturing device performs an iterative calculation based on the Mth-layer calculation result and the CNN when M is less than N and the threshold time is not reached.

6. A recognition method, comprising:

by an image capturing device, generating a Mth-layer calculation result based on image data and a convolutional neural network (CNN);

by the image capturing device, transmitting feature information associated with the Mth-layer calculation result to a server, wherein M is a positive integer, M is equal to or greater than 1, M is less than or equal to N, and N is a predetermined positive integer;

by the server, generating a Kth-layer calculation result based on the feature information and the CNN by an iterative method when M is less than N, wherein K is a positive integer which is greater than M and less than or equal to N; and

by the server, generating a first recognition result associated with the image data based on the Kth-layer calculation result and a first recognition model when K is equal to N, in order to recognize the image data.

7. The recognition method of claim 6 , wherein when M is equal to N, the Mth-layer calculation result is configured to be as the feature information and the recognition method further comprises:

by the server, generating the first recognition result based on the feature information and the first recognition model when M is equal to N.

8. The recognition method of claim 6 , further comprising:

by the server, generating a second recognition result associated with the image data based on the first recognition result and a second recognition model,

wherein the second recognition model is different from the first recognition model.

9. The recognition method of claim 6 , wherein the Mth-layer calculation result comprises a convolution result, a pooling result, an activation result, or a deconvolution result.

10. The recognition method of claim 6 , further comprising:

by the image capturing device, determining whether a threshold time is reached or not;

by the image capturing device, transmitting the Mth-layer calculation result to the server to be as the feature information when M is less than N and the threshold time is reached; and

by the image capturing device, performing an iterative calculation based on the Mth-layer calculation result and the CNN when M is less than N and the threshold time is not reached.

11. A non-transitory computer readable storage medium storing a computer program, wherein the computer program is configured to execute a recognition method, and the recognition method comprises:

generating a Mth-layer calculation result based on image data and a convolutional neural network (CNN);

transmitting feature information associated with the Mth-layer calculation result, wherein M is a positive integer, M is equal to or greater than 1, M is less than or equal to N, and N is a predetermined positive integer;

generating a Kth-layer calculation result based on the feature information and the CNN by an iterative method when M is less than N, wherein K is a positive integer which is greater than M and less than or equal to N; and

generating a first recognition result associated with the image data based on the Kth-layer calculation result and a first recognition model when K is equal to N, in order to recognize the image data.