Cross-Platform Neural Video Coding: A Case Study
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866916456032108544 |
|---|---|
| author | Conceição, Ruhan Porto, Marcelo Peng, Wen-Hsiao Agostini, Luciano |
| author_facet | Conceição, Ruhan Porto, Marcelo Peng, Wen-Hsiao Agostini, Luciano |
| contents | In this paper, we first show that current learning-based video codecs, specifically the SSF codec, are not suitable for real-world applications due to the mismatch between the encoder and decoder caused by floating-point round-off errors. To address this issue, we propose the static quantization of the hyper prior decoding path. The quantization parameters are determined through an exhaustive search of all possible combinations of observers and quantization schemes from PyTorch. For the SSF codec, when encoding and decoding on different machines, the proposed solution effectively mitigates the mismatch issue and enhances compression efficiency results by preventing severe image quality degradation. When encoding and decoding are performed on the same machine, it constrains the average BD-rate increase to 9.93% and 9.02% for UVG and HEVC-B sequences, respectively. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2410_20145 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Cross-Platform Neural Video Coding: A Case Study Conceição, Ruhan Porto, Marcelo Peng, Wen-Hsiao Agostini, Luciano Multimedia In this paper, we first show that current learning-based video codecs, specifically the SSF codec, are not suitable for real-world applications due to the mismatch between the encoder and decoder caused by floating-point round-off errors. To address this issue, we propose the static quantization of the hyper prior decoding path. The quantization parameters are determined through an exhaustive search of all possible combinations of observers and quantization schemes from PyTorch. For the SSF codec, when encoding and decoding on different machines, the proposed solution effectively mitigates the mismatch issue and enhances compression efficiency results by preventing severe image quality degradation. When encoding and decoding are performed on the same machine, it constrains the average BD-rate increase to 9.93% and 9.02% for UVG and HEVC-B sequences, respectively. |
| title | Cross-Platform Neural Video Coding: A Case Study |
| topic | Multimedia |
| url | https://arxiv.org/abs/2410.20145 |