DSSLIC: Deep Semantic Segmentation-based Layered Image Compression
Deep learning has revolutionized many computer vision fields in the last few years, including learning-based image compression. In this paper, we propose a deep semantic segmentation-based layered image compression (DSSLIC) framework in which the semantic segmentation map of the input image is obtained and encoded as a part of the bit-stream. A compact representation of the input image is also generated and encoded as the base layer. The segmentation map and the compact version of the image are then employed to obtain a coarse reconstruction of the image. As an enhancement layer in the bit-stream, the residual between the input and the coarse reconstruction is additionally encoded. Experimental results show that the proposed framework can outperform the H.265/HEVC-based BPG codec and other codecs at low bit rates in both PSNR and MS-SSIM metrics. Besides, since semantic segmentation map is included in the bit-stream, the proposed scheme can facilitate many other tasks such as image search and object-based adaptive image compression.
READ FULL TEXT