AI Paper Atlas

Research map / Detection and segmentation / CNN features

Representing Topological Self-Similarity Using Fractal Feature Maps for Accurate Segmentation of Tubular Structures

Jiaxing Huang, Yanfeng Zhou, Yaoru Luo, Guole Liu, Heng Guo, Ge Yang

ECCV 2024 · Detection and segmentation › CNN features · approach: Context modeling for semantic segmentation · 16 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add FFM_image as extra input channel to U-NetU-Net · ROADIoU: 62.47 -> 65.74Improved
add FFM_image as extra input channel to HR-NetHR-Net · ROADIoU: 62.89 -> 65.60Improved
add FFM_image to U-Net on non-tubular NUCLEUSU-Net · NUCLEUSIoU: 78.02 -> 78.46Improved
add FFM_image channel rather than duplicating the image channel (ablation)U-Net · ERIoU: 75.61 -> 76.59Improved
use FFM_image instead of Hurst/mean/contrast features as extra input (ablation)U-Net · STAREIoU: 66.83 -> 68.07Improved
add edge and skeleton decoders to U-Net (no FFM) (ablation)U-Net · ERIoU: 75.44 -> 76.28Improved
add skeleton decoder to MD-Net with FFM and edge decoder (ablation)MD-Net · ERIoU: 76.97 -> 77.09Improved
add FFM_label as loss weight (L_constrained vs L_global)MD-Net · ERIoU: 77.01 -> 77.09Mixed
vary FFM window size (5 to 11) and step size (1 to 4) (ablation)U-Net and MD-Net · ER and STAREIoU: variation within ~0.3%No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy