ConvNeXt-ChARM：用于高效神经图像压缩的基于 ConvNeXt 的变换
ConvNeXt-ChARM: ConvNeXt-based Transform for Efficient Neural Image Compression
AI总结：
提出基于 ConvNeXt 与通道自回归先验的高效图像压缩框架 ConvNeXt-ChARM，在四个数据集上以更低计算成本取得优于 VTM-18.0 和 SwinT-ChARM 的压缩性能。
摘要：
在过去几年中，神经图像压缩受到了研究界和工业界的广泛关注，并产生了有前景的端到端深度神经编解码器，其率失真性能优于传统编解码器。尽管取得了显著进展，但当前方法（包括基于注意力的变换编码）在保持重建保真度的同时降低编码码率方面仍有待改进，尤其是在非均匀纹理图像区域。这些模型还需要更多参数和更长解码时间。为应对上述挑战，我们提出 ConvNeXt-ChARM，这是一种高效的基于 ConvNeXt 的变换编码框架，并结合计算高效的通道自回归先验，以从超先验潜表示和量化潜表示中捕获全局与局部上下文。所提架构可进行端到端优化，以充分利用上下文信息并提取紧凑的潜表示，同时重建更高质量的图像。在四个广泛使用的数据集上的实验结果表明，与多功能视频编码（VVC）参考编码器（VTM-18.0）和最先进的学习型图像压缩方法 SwinT-ChARM 相比，ConvNeXt-ChARM 分别带来平均 5.24% 和 1.22% 的持续且显著的 BD-rate（PSNR）降低。此外，我们提供模型缩放研究以验证方法的计算效率，并开展多项客观和主观分析，以凸显下一代 ConvNet（即 ConvNeXt）与 Swin Transformer 之间的性能差距。
英文摘要：
Over the last few years, neural image compression has gained wide attention from research and industry, yielding promising end-to-end deep neural codecs outperforming their conventional counterparts in rate-distortion performance. Despite significant advancement, current methods, including attention-based transform coding, still need to be improved in reducing the coding rate while preserving the reconstruction fidelity, especially in non-homogeneous textured image areas. Those models also require more parameters and a higher decoding time. To tackle the above challenges, we propose ConvNeXt-ChARM, an efficient ConvNeXt-based transform coding framework, paired with a compute-efficient channel-wise auto-regressive prior to capturing both global and local contexts from the hyper and quantized latent representations. The proposed architecture can be optimized end-to-end to fully exploit the context information and extract compact latent representation while reconstructing higher-quality images. Experimental results on four widely-used datasets showed that ConvNeXt-ChARM brings consistent and significant BD-rate (PSNR) reductions estimated on average to 5.24% and 1.22% over the versatile video coding (VVC) reference encoder (VTM-18.0) and the state-of-the-art learned image compression method SwinT-ChARM, respectively. Moreover, we provide model scaling studies to verify the computational efficiency of our approach and conduct several objective and subjective analyses to bring to the fore the performance gap between the next generation ConvNet, namely ConvNeXt, and Swin Transformer.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
