Conditional Encoder-Based Adaptive Deep Image Compression with Classification-Driven Semantic Awareness
This paper proposes a new algorithm for adaptive deep image compression (DIC) that can compress images for different purposes or contexts at different rates. The algorithm can compress images with semantic awareness, which means classification-related semantic features are better protected in lossy...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-06-01
|
| Series: | Electronics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2079-9292/12/13/2781 |
| _version_ | 1797591930183352320 |
|---|---|
| author | Zhongyue Lei Weicheng Zhang Xuemin Hong Jianghong Shi Minxian Su Chaoheng Lin |
| author_facet | Zhongyue Lei Weicheng Zhang Xuemin Hong Jianghong Shi Minxian Su Chaoheng Lin |
| author_sort | Zhongyue Lei |
| collection | DOAJ |
| description | This paper proposes a new algorithm for adaptive deep image compression (DIC) that can compress images for different purposes or contexts at different rates. The algorithm can compress images with semantic awareness, which means classification-related semantic features are better protected in lossy image compression. It builds on the existing conditional encoder-based DIC method and adds two features: a model-based rate-distortion-classification-perception (RDCP) framework to control the trade-off between rate and performance for different contexts, and a mechanism to generate coding conditions based on image complexity and semantic importance. The algorithm outperforms the QMAP2021 benchmark on the ImageNet dataset. Over the tested rate range, it improves the classification accuracy by <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>11</mn><mo>%</mo></mrow></semantics></math></inline-formula> and the perceptual quality by <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>12.4</mn><mo>%</mo></mrow></semantics></math></inline-formula>, <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>32</mn><mo>%</mo></mrow></semantics></math></inline-formula>, and <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>1.3</mn><mo>%</mo></mrow></semantics></math></inline-formula> on average for NIQE, LPIPS, and FSIM metrics, respectively. |
| first_indexed | 2024-03-11T01:44:34Z |
| format | Article |
| id | doaj.art-cf8826d95432471d940fbdbc432403e7 |
| institution | Directory Open Access Journal |
| issn | 2079-9292 |
| language | English |
| last_indexed | 2024-03-11T01:44:34Z |
| publishDate | 2023-06-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Electronics |
| spelling | doaj.art-cf8826d95432471d940fbdbc432403e72023-11-18T16:23:22ZengMDPI AGElectronics2079-92922023-06-011213278110.3390/electronics12132781Conditional Encoder-Based Adaptive Deep Image Compression with Classification-Driven Semantic AwarenessZhongyue Lei0Weicheng Zhang1Xuemin Hong2Jianghong Shi3Minxian Su4Chaoheng Lin5School of Informatics, Xiamen University, Xiamen 361005, ChinaSchool of Informatics, Xiamen University, Xiamen 361005, ChinaSchool of Informatics, Xiamen University, Xiamen 361005, ChinaSchool of Informatics, Xiamen University, Xiamen 361005, ChinaXiamen Satellite Positioning Application Co., Ltd., Xiamen 361008, ChinaXiamen Beidou Key Laboratory of Applied Technology, Xiamen 361008, ChinaThis paper proposes a new algorithm for adaptive deep image compression (DIC) that can compress images for different purposes or contexts at different rates. The algorithm can compress images with semantic awareness, which means classification-related semantic features are better protected in lossy image compression. It builds on the existing conditional encoder-based DIC method and adds two features: a model-based rate-distortion-classification-perception (RDCP) framework to control the trade-off between rate and performance for different contexts, and a mechanism to generate coding conditions based on image complexity and semantic importance. The algorithm outperforms the QMAP2021 benchmark on the ImageNet dataset. Over the tested rate range, it improves the classification accuracy by <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>11</mn><mo>%</mo></mrow></semantics></math></inline-formula> and the perceptual quality by <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>12.4</mn><mo>%</mo></mrow></semantics></math></inline-formula>, <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>32</mn><mo>%</mo></mrow></semantics></math></inline-formula>, and <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>1.3</mn><mo>%</mo></mrow></semantics></math></inline-formula> on average for NIQE, LPIPS, and FSIM metrics, respectively.https://www.mdpi.com/2079-9292/12/13/2781deep image codingimage semanticadaptive codinghybrid contexts |
| spellingShingle | Zhongyue Lei Weicheng Zhang Xuemin Hong Jianghong Shi Minxian Su Chaoheng Lin Conditional Encoder-Based Adaptive Deep Image Compression with Classification-Driven Semantic Awareness Electronics deep image coding image semantic adaptive coding hybrid contexts |
| title | Conditional Encoder-Based Adaptive Deep Image Compression with Classification-Driven Semantic Awareness |
| title_full | Conditional Encoder-Based Adaptive Deep Image Compression with Classification-Driven Semantic Awareness |
| title_fullStr | Conditional Encoder-Based Adaptive Deep Image Compression with Classification-Driven Semantic Awareness |
| title_full_unstemmed | Conditional Encoder-Based Adaptive Deep Image Compression with Classification-Driven Semantic Awareness |
| title_short | Conditional Encoder-Based Adaptive Deep Image Compression with Classification-Driven Semantic Awareness |
| title_sort | conditional encoder based adaptive deep image compression with classification driven semantic awareness |
| topic | deep image coding image semantic adaptive coding hybrid contexts |
| url | https://www.mdpi.com/2079-9292/12/13/2781 |
| work_keys_str_mv | AT zhongyuelei conditionalencoderbasedadaptivedeepimagecompressionwithclassificationdrivensemanticawareness AT weichengzhang conditionalencoderbasedadaptivedeepimagecompressionwithclassificationdrivensemanticawareness AT xueminhong conditionalencoderbasedadaptivedeepimagecompressionwithclassificationdrivensemanticawareness AT jianghongshi conditionalencoderbasedadaptivedeepimagecompressionwithclassificationdrivensemanticawareness AT minxiansu conditionalencoderbasedadaptivedeepimagecompressionwithclassificationdrivensemanticawareness AT chaohenglin conditionalencoderbasedadaptivedeepimagecompressionwithclassificationdrivensemanticawareness |