Method and device for performing artificial intelligence encoding and artificial intelligence decoding

1 . An artificial intelligence (AI) encoding apparatus comprising:

a memory configured to store instructions; and

at least one processor configured to execute the instructions to:

obtain an original image, previously-encoded frame information, and network environment information;

obtain deblocking filter setting information, based on the original image, the previously-encoded frame information, and the network environment information;

perform deblocking filtering to the original image, based on the deblocking filter setting information to obtain a deblocking-filtered original image;

obtain an AI-downscaled first image by providing the deblocking-filtered original image to a downscaling deep neural network (DNN);

obtain a quantization parameter differential value map and information about rate control based on the original image, the previously-encoded frame information, and the network environment information;

generate image data by performing first encoding on the AI-downscaled first image based on the quantization parameter differential value map and the information about the rate control; and

transmit the deblocking filter setting information, AI data including information related to the AI downscaling, and the image data.

2 . The AI encoding apparatus of claim 1 , wherein the deblocking filter setting information comprises a blocking area map which represents areas where blocking artifacts occur.

3 . The AI encoding apparatus of claim 2 , wherein the blocking area map is determined based on at least one of a complexity of the original image and a movement change degree of the original image.

4 . The AI encoding apparatus of claim 1 , wherein the deblocking filtering is used to redistribute a bit allotment, based on the deblocking filter setting information.

5 . The AI encoding apparatus of claim 1 , wherein DNN setting information of the downscaling DNN is obtained based on joint training between the downscaling DNN and an upscaling DNN for performing AI upscaling based on the image data.

6 . The AI encoding apparatus of claim 1 , wherein, based on the network environment information indicating that there is no change in a network environment, the deblocking filter setting information is not obtained, and the deblocking filtering is not performed.

7 . The AI encoding apparatus of claim 1 , wherein the network environment information indicates a change in at least one of a bitrate of a network channel and a bandwidth of the network channel.

8 . The AI encoding apparatus of claim 1 , wherein the previously-encoded frame information indicates whether a blocking artifact of an encoded previous frame is detected.

9 . An artificial intelligence (AI) encoding method performed by an AI encoding apparatus, the AI encoding method comprising:

obtaining an original image, previously-encoded frame information, and network environment information;

obtaining deblocking filter setting information, based on the original image, the previously-encoded frame information, and the network environment information;

applying deblocking filtering to the original image, based on the deblocking filter setting information, to obtain a deblocking-filtered original image;

obtaining an AI-downscaled first image by providing the deblocking-filtered original image to a downscaling deep neural network (DNN);

obtain a quantization parameter differential value map and information about rate control based on the original image, the previously-encoded frame information, and the network environment information;

generating image data by performing first encoding on the AI-downscaled first image based on the quantization parameter differential value map and the information about the rate control; and

transmitting the deblocking filter setting information, AI data including information related to the AI downscaling, and the image data.