영스퀘어

1) Paper Bibliography Title : Video Super-Resolution With Convolutional Neural Networks Authors : Armin Kappeler, Seunghwan Yoo, Qiqin Dai, and Aggelos K. Katsaggelos IEEE transactions on computational imaging June, 2016 2) Problems & Motivations Problems Most video content is not available in UHD resolution. Therefore SR algorithms are needed to generate UHD content from Full HD (FHD) (1,920 × ..

CUDA 11.1 Pytorch 1.8.0 GCN Layer 사이에 Attention mechanism을 구현하는 도중 아래와 같은 에러가 발생했다. RuntimeError: one of the variables needed for gradient computation has been modified by an inplace operation: [torch.cuda.FloatTensor [32]], which is output 0 of SelectBackward, is at version 165; expected version 164 instead. Hint: enable anomaly detection to find the operation that failed to compute its gradien..

결과 결과적으로, Distortion 값이 Perspective < Affine < Translation 순이라는 것을 확인 할 수 있다. 이를 통해, Motion model의 차이가 압축 성능에 크게 영향을 미침을 예상할 수 있다.

소스 코드 using System; using System.IO; namespace MEModule_Perspective { class Program { static void Main(string[] args) { ... } private static byte[,,] full_search(byte[] luma, int frame, int width, int height) { byte[,,] result = new byte[frame - 1, width, height]; ulong[] allResi = new ulong[frame - 1]; int MB = 32; // 매크로블록 사이즈 int SubMB = 16; // Sub 블록 사이즈 int SU = 16; // 매크로블록에서 위아래 얼마나 더 가져올..

소스 코드 using System; using System.IO; namespace MEModule_Affine { class Program { static void Main(string[] args) { ... byte[] result_stream = make_stream(full_search(luma, frame, width, height), frame, width, height); ... } private static byte[,,] full_search(byte[] luma, int frame, int width, int height) { byte[,,] result = new byte[frame - 1, width, height]; ulong[] allResi = new ulong[frame -..

사전 정의 소스 코드 using System; using System.IO; namespace MEModule_ORG { class Program { static void Main(string[] args) { string path = "../../../sequence/BasketballDrive_1920x1080_50.yuv"; FileStream fs = new FileStream(path, FileMode.Open); BinaryReader br = new BinaryReader(fs); //Sequence size, number of frames int width = 1920; int height = 1080; int frame = 10; int N = width * height; byte[] l..

[References] Li, Tianyi, Mai Xu, and Xin Deng. "A deep convolutional neural network approach for complexity reduction on intra-mode HEVC." 2017 IEEE International Conference on Multimedia and Expo (ICME). IEEE, 2017. JEM reference software, https://jvet.hhi.fraunhofer.de/svn/svn_HMJEMSoftware/ J. Chen, E. Alshina, G. J. Sullivan, J.-R. Ohm, and J. Boyce, “Algorithm Description of Joint Explorati..

[References] Li, Tianyi, Mai Xu, and Xin Deng. "A deep convolutional neural network approach for complexity reduction on intra-mode HEVC." 2017 IEEE International Conference on Multimedia and Expo (ICME). IEEE, 2017.

[References] Li, Tianyi, Mai Xu, and Xin Deng. "A deep convolutional neural network approach for complexity reduction on intra-mode HEVC." 2017 IEEE International Conference on Multimedia and Expo (ICME). IEEE, 2017.

[References] Shuai Huo, Dong Liu, Feng Wu, Houqiang Li, “Convolutional Neural Network-Based Motion Compensation Refinement for Video Coding,” in IEEE International Symposium on Circuits and Systems (ISCAS), 2018, pp. 1–4. Zhenghui Zhao, Shiqi Wang, Shanshe Wang, Xinfeng Zhang, Siwei Ma, Jiansheng Yang, “CNN-Based Bi-Directional Motion Compensation for High Efficiency Video Coding,” in IEEE Inter..