Methods and apparatuses for performing encoding and decoding on image

1. A server for providing an image by using artificial intelligence (AI), the server comprising:

one or more processors configured to execute one or more instructions stored in the server to:

select a down-scaling deep neural network (DNN) setting information among a plurality of down-scaling DNN setting information for AI down-scaling an original image of at least one frame,

obtain a down-scaled image of the at least one frame by performing the AI down-scaling of the original image of the at least one frame through a down-scaling DNN which is set with the selected down-scaling DNN setting information, and

obtain AI data related to the AI down-scaling and obtain image data by encoding the down-scaled image of the at least one frame, to obtain a video file including a media data box which comprises the image data and the AI data,

wherein the AI data is included in supplementary enhancement information (SEI) message which is a data unit including additional information about an image, the AI data indicating whether the AI up-scaling is to be performed or the AI up-scaling is not to be performed on the at least one frame.

2. The server of claim 1 , wherein the AI data includes an index indicating DNN setting information for down-scaling among a plurality of DNN setting information for down-scaling, information related to at least one of a resolution difference between the original image and the down-scaled image, a bitrate regarding the image data, a quantization parameter regarding the image data, a resolution of the down-scaled image, or a codec type used to encode the down-scaled image.

3. An electronic device for displaying an image by using an artificial intelligence (AI), the electronic device comprising:

a display; and

one or more processors configured to execute one or more instructions stored in the electronic device to:

receive a video file including image data and AI data about AI up-scaling of the image data which includes AI data indicating whether the AI up-scaling is performed or the AI up-scaling is not to be performed on at least one frame,

obtain the AI data from supplementary enhancement information (SEI) message, which is a data unit including additional information about an image, in a media data box of the video file and obtain the image data from the media data box of the video file, reconstruct a down-scaled image of at least one frame by decoding the image data, obtain up-scaling deep neural network (DNN) setting information of the at least one frame used to perform the AI up-scaling on the at least one frame through an up-scaling deep neural network selected from among a plurality of DNNs related to a plurality of up-scaling DNN setting information based on the AI data, when the AI up-scaling is to be performed on the at least one frame,

generate an AI up-scaled image corresponding to the down-scaled image by performing the AI up-scaling of the down-scaled image through the selected up-scaling deep neural network, and

provide on the display of the electronic device, the AI up-scaled image.

4. The electronic device of claim 3 , wherein the AI data includes an index indicating DNN setting information for down-scaling among a plurality of DNN setting information for down-scaling, information related to at least one of a resolution difference between the original image and the down-scaled image, a bitrate regarding the image data, a quantization parameter regarding the image data, a resolution of the down-scaled image, or a codec type used to encode the down-scaled image.

5. A non-transitory computer-readable recording medium having stored thereon a video file,

wherein the video file comprises a media data box comprising:

image data including encoding information of a down-scaled image generated by AI down-scaling an original image; and

AI data about AI up-scaling of the down-scaled image to be reconstructed according to the image data,

wherein the AI data is included in supplementary enhancement information (SEI) message which is a data unit including additional information about an image,

wherein the AI data comprises:

AI data indicating whether the AI up-scaling is to be performed or the AI up-scaling is not to be performed on at least one frame; and

AI data about up-scaling deep neural network (DNN) setting information used for performing the AI up-scaling of the at least one frame through an up-scaling deep neural network selected from among a plurality of DNNs related to a plurality of DNN setting information, when the AI up-scaling is to be performed on the at least one frame.

6. The non-transitory computer-readable recording medium of claim 5 , wherein the AI data, indicating whether the AI up-scaling is to performed or the AI up-scaling is not to be performed, comprises at least one of:

video AI data indicating whether the AI up-scaling is to be performed on a plurality of frames included in the image data;

video segment AI data indicating whether the AI up-scaling is to be performed on a plurality of frames included in a video segment;

frame group AI data indicating whether the AI up-scaling is to be performed on a plurality of frames included in a frame group; or frame AI data indicating whether the AI up-scaling is to be performed on a current frame.

7. The non-transitory computer-readable recording medium of claim 5 , wherein the AI data about the up-scaling deep neural network (DNN) setting information comprises at least one of:

video AI data about at least one piece of up-scaling DNN setting information used to perform the AI up-scaling of all of a plurality of frames included in the image data;

video segment AI data about at least one piece of up-scaling DNN setting information used to perform the AI up-scaling of the plurality of frames included in a video segment;

frame group AI data about at least one piece of up-scaling DNN setting information used to perform the AI up-scaling of the plurality of frames included in a frame group; or

frame AI data about up-scaling DNN setting information used to perform the AI up-scaling of a current frame.

8. The non-transitory computer-readable recording medium of claim 7 , wherein:

the AI data comprises at least one of:

video segment AI data dependency information indicating whether video segment AI data about at least one piece of up-scaling DNN setting information is the same between a current video segment and a consecutive previous video segment,

frame group AI data dependency information indicating whether frame group AI data about at least one piece of up-scaling DNN setting information is the same between a current frame group and a consecutive previous frame group, or frame AI data dependency information indicating whether frame AI data about at least one piece of up-scaling DNN setting information is the same between the current frame and a consecutive previous frame,

when the video segment AI data dependency information indicates that the AI data about at least one piece of up-scaling DNN setting information is the same between the current video segment and the consecutive previous video segment, the video segment AI data about at least one piece of up-scaling DNN setting information about the current video segment is omitted from the AI data,

when the frame group AI data dependency information indicates that the frame group AI data about at least one piece of up-scaling DNN setting information is the same between the current frame group and the consecutive previous frame group, the frame group AI data about at least one piece of up-scaling DNN setting information about the current frame group is omitted from the AI data, and

when the frame AI data dependency information indicates that the frame AI data about at least one piece of up-scaling DNN setting information is the same between the current frame and the consecutive previous frame, the frame AI data about at least one of up-scaling DNN setting information about the current frame is omitted from the AI data.

9. The non-transitory computer-readable recording medium of claim 5 , wherein the AI data about at least one piece of up-scaling DNN setting information comprises at least one of:

channel information indicating a color channel to which the AI up-scaling is performed;

target bitrate information indicating a bitrate of the down-scaled image; or resolution information related to resolution of an AI up-scaled image or the original image.