Lean Unet: 一种用于图像分割的紧凑模型
Lean Unet: A Compact Model for Image Segmentation
AI总结：
本文提出了一种紧凑的U-net架构（LUnet），通过减少参数数量和优化结构，实现了与传统U-net和自适应剪枝网络相当的性能。
摘要：
U-net及其变种已成为语义图像分割的标准方法，尤其是在计算机辅助放射学领域。当前的U-net架构通过迭代降低空间分辨率同时增加通道维度来保持信息内容。这种结构需要较大的内存占用，限制了训练批次大小并增加推理延迟。通道剪枝可以在不损失精度的情况下压缩U-net架构，但需要长时间的优化，并且可能无法在不同任务和数据集上通用。通过研究U-net剪枝，我们假设最终结构是关键因素，而不是剪枝策略中的通道选择。基于我们的观察，我们提出了一种紧凑的U-net架构（LUnet），其具有紧凑且扁平的层次结构，其中通道在分辨率减半时不会翻倍。我们在一个公开的MRI数据集上进行了评估，允许进行可比的报告，以及在两个内部CT数据集上。我们证明，最先进的剪枝解决方案（STAMP）主要从具有最高通道数的层进行剪枝。相比之下，简单地在剪枝识别的层或在最大层上随机消除一个通道可以实现相似或更好的性能。我们提出的LUnet具有固定架构和超过30倍更少的参数，其性能与传统U-net对手和数据自适应剪枝网络相当。所提出的紧凑U-net在各层中保持通道数恒定，需要远少的参数，同时在相同总参数数量下实现了优于标准U-net的性能。跳跃连接允许U-net瓶颈通道大幅减少，与需要增加瓶颈通道以传播信息的标准编码器-解码器架构不同。
英文摘要：
Unet and its variations have been standard in semantic image segmentation, especially for computer assisted radiology. Current Unet architectures iteratively downsample spatial resolution while increasing channel dimensions to preserve information content. Such a structure demands a large memory footprint, limiting training batch sizes and increasing inference latency. Channel pruning compresses Unet architecture without accuracy loss, but requires lengthy optimization and may not generalize across tasks and datasets. By investigating Unet pruning, we hypothesize that the final structure is the crucial factor, not the channel selection strategy of pruning. Based on our observations, we propose a lean Unet architecture (LUnet) with a compact, flat hierarchy where channels are not doubled as resolution is halved. We evaluate on a public MRI dataset allowing comparable reporting, as well as on two internal CT datasets. We show that a state-of-the-art pruning solution (STAMP) mainly prunes from the layers with the highest number of channels. Comparatively, simply eliminating a random channel at the pruning-identified layer or at the largest layer achieves similar or better performance. Our proposed LUnet with fixed architectures and over 30 times fewer parameters achieves performance comparable to both conventional Unet counterparts and data-adaptively pruned networks. The proposed lean Unet with constant channel count across layers requires far fewer parameters while achieving performance superior to standard Unet for the same total number of parameters. Skip connections allow Unet bottleneck channels to be largely reduced, unlike standard encoder-decoder architectures requiring increased bottleneck channels for information propagation.
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。
