Depth Map Prediction for Fast Block Partitioning in HEVC
The project page for the paper:
Aolin Feng, Changsheng Gao, Li Li, Dong Liu, and Feng Wu."CNN-based Depth Map Prediction for Fast Block Partitioning in HEVC Intra Coding", IEEE International Conference on Multimedia and Expo (ICME), 2021. OpenAccess
@inproceedings{feng2021cnn,
title={Cnn-Based Depth Map Prediction for Fast Block Partitioning in HEVC Intra Coding},
author={Feng, Aolin and Gao, Changsheng and Li, Li and Liu, Dong and Wu, Feng},
booktitle={2021 IEEE International Conference on Multimedia and Expo (ICME)},
pages={1--6},
year={2021},
organization={IEEE}
}
- Dependency: Pytroch, GPU (the code only can be run in GPU mode)
- Command: python dp_total_test.py > test.log
- cfg: Encoding configuration file
- cfg/per-sequence: Sequence information file
- Test_Sequence_List.txt: Names of sequences to be tested, which correspond to the names of sequence information file
- codec: Encoder.exe + Decoder.exe
- models: Trained models for depth map prediction
- DepthFlag: Output split flags used for encoding acceleration
- output: Output bit stream
- log: Output encoding and decoding logs
- The working directory mus be the current directory, and the folder names cannot be modified.
- To test a sequence:
- Write the sequence information file strictly according to the format of existing files. Use '/' rather than '\' in the file path.
- Write all the names of test sequences in "Test_Sequence_List.txt".
- In the output log file of the main python script, the last line stores the network inference time, which shpuld be added to the total encoding time.