Electronic device and method for controlling thereof

1. An electronic device, comprising:

a memory comprising a neural network model trained to identify private information of a user of the electronic device comprised in an image;

a display;

a communicator comprising circuitry; and

a processor configured to:

based on a user command to transmit a first image to an external device being input, identify whether the private information of the user is comprised in the first image by inputting the first image in the neural network model,

based on identifying that the private information of the user is comprised in the first image, control the display to display a first user interface (UI) asking whether to process at least one piece of the private information based on the private information comprised in the first image, and

based on a user command input through the first UI, process the first image and control the communicator to transmit the processed first image to the external device,

wherein the neural network model comprises:

an object detection model trained to detect one or more objects comprised in the first image and classify a detected object as at least one of text or an image;

a text classification model trained to output a first privacy level corresponding to an object classified as the text from among detected objects; and

an image classification model trained to output a second privacy level corresponding to an object classified as the image from among the detected objects,

wherein the processor is further configured to:

based on the first privacy level exceeding a first threshold value or the second privacy level exceeding a second threshold value which is different from the first threshold value, identify that the private information of the user is comprised in the first image, and

obtain meta data on the private information comprised in the first image through the neural network model, and

wherein the text classification model comprises a text type classification model which is trained to output the first privacy level corresponding to a text type based on the text type obtained through a rule-based classification model and final feature data obtained through a feature data processing model.

2. The electronic device of claim 1 , wherein the text classification model is characterized by:

identifying a type corresponding to the object classified as the text; and

applying a pre-defined weight to the identified type to output data to calculate the first privacy level.

3. The electronic device of claim 1 , wherein the processor is further configured to:

use information on a location of the private information comprised in the first image from among the meta data, and

control the display to display the first UI in which a bounding box, which shows an area in which the private information is located from among the first image, is comprised.

4. The electronic device of claim 1 , wherein the processor is further configured to:

based on a user command of not processing the private information comprised in the first image being input through the first UI, control the communicator to transmit the first image to the external device, and

based on a user command of processing the private information comprised in the first image being input through the first UI, perform a blurring process or a masking process of an area in which the private information is comprised in the first image and control the communicator to transmit the processed first image to the external device.

5. The electronic device of claim 1 , wherein the processor is further configured to, based on identifying that the private information of the user is not comprised in the first image, control the communicator to transmit the first image to the external device.

6. The electronic device of claim 1 , further comprising:

a camera;

wherein the processor is further configured to:

based on a second image being obtained by capturing surroundings of the electronic device through the camera, identify whether the private information of the user in comprised in the second image by inputting the second image in the neural network model, and

based on the private information of the user being comprised in the second image, control the display to display a second UI requesting a user authentication in response to a user command selecting the second image being input.

7. The electronic device of claim 1 , wherein the processor is further configured to, based on the external device being a social networking service (SNS) server managing an SNS, upload an image in which an area comprising the private information of the user is processed from among the first image based on meta data in an SNS account of the user by controlling the communicator to transmit the first image and meta data on the private information comprised in the first image.

8. The electronic device of claim 1 , wherein the processor is further configured to:

based on identifying whether the private information of the user is comprised in the first image through the neural network model, control the display to display a third UI asking whether the private information not identified by the neural network model is comprised in the first image, and

additionally train the neural network model based on information input from the user through the third UI.

9. The electronic device of claim 1 , wherein the text type classification model is trained to obtain a calculation value by applying pre-defined weights matched for respective types of text for respective types of text and output the first privacy level output by using the obtained calculation value and the final feature data.

10. A method of controlling an electronic device comprising a neural network model trained to identify private information of a user of the electronic device comprised in an image, the method comprising:

based on a user command to transmit a first image to an external device being input, identifying whether the private information of the user is comprised in the first image by inputting the first image in the neural network model;

based on identifying that the private information of the user is comprised in the first image, displaying a first user interface (UI) asking whether to process at least one piece of the private information based on the private information comprised in the first image; and

based on a user command input through the first UI, processing the first image and transmitting the processed first image to the external devices,

wherein the neural network model comprises:

an object detection model trained to detect one or more objects comprised in the first image and classify a detected object as at least one of text or an image;

a text classification model trained to output a first privacy level corresponding to an object classified as the text from among detected objects; and

an image classification model trained to output a second privacy level corresponding to an object classified as the image from among the detected objects,

wherein the identifying comprises:

based on the first privacy level exceeding a first threshold value or the second privacy level exceeding a second threshold value which is different from the first threshold value, identifying that the private information of the user is comprised in the first image; and

obtaining meta data on the private information comprised in the first image through the neural network model, and

wherein the text classification model comprises a text type classification model which is trained to output the first privacy level corresponding to a text type based on the text type obtained through a rule-based classification model and final feature data obtained through a feature data processing model.

11. The method of claim 10 , wherein the text classification model is characterized by:

identifying a type corresponding to the object classified as the text; and

applying a pre-defined weight to the identified type to output data to calculate the first privacy level.

12. The method of claim 10 , wherein the displaying of the first UI comprises:

using information on a location of the private information comprised in the first image from among the meta data; and

displaying the first UI in which a bounding box, which shows an area in which the private information is located from among the first image, is comprised.