Scholarlink

<p>The significance of multi-scale features has been gradually recognized by the edge detection community. However, the fusion of multi-scale features increases the complexity of the model, which is not friendly to practical application. In this work, we propose a Compact Twice Fusion Network (CTFN) to fully integrate multi-scale features while maintaining the compactness of the model. CTFN includes two lightweight feature fusion modules: a Semantic Enhancement Module (SEM) that can utilize the semantic information contained in coarse-scale features to guide the learning of fine-scale features, and a Pseudo Pixel-level Weighting (PPW) module that aggregates the complementary merits of features by assigning weights to all features. Furthermore, we propose a novel loss function, coined Dynamic Focal Loss, which reshapes the standard cross-entropy loss and dynamically adjusts the weights to correct the distribution of hard samples. We evaluate our method on three datasets, i.e., BSDS500, NYUDv2, and BIPEDv2. CTFN achieves competitive accuracy with fewer parameters and computational costs than state-of-the-art methods. With the exception of the backbone, CTFN only requires 0.1M extra parameters, making the computation 40% less expensive than other state-of-the-art methods. The codes are available at <a href="https://github.com/Li-yachuan/CTFN-pytorch-master">https://github.com/Li-yachuan/CTFN-pytorch-master</a>.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Compact twice fusion network for edge detection

  • 期刊论文

  • 发表日期：

    2025年1月7日

查看全文
  • Zongmin Li,
  • Yachuan Li,
  • P. Xavier Soria,
  • Chaozhi Yang,
  • Qian Xiao,
  • Yun Bai,
  • Hua Li,
  • Xiangdong Wang

摘要

The significance of multi-scale features has been gradually recognized by the edge detection community. However, the fusion of multi-scale features increases the complexity of the model, which is not friendly to practical application. In this work, we propose a Compact Twice Fusion Network (CTFN) to fully integrate multi-scale features while maintaining the compactness of the model. CTFN includes two lightweight feature fusion modules: a Semantic Enhancement Module (SEM) that can utilize the semantic information contained in coarse-scale features to guide the learning of fine-scale features, and a Pseudo Pixel-level Weighting (PPW) module that aggregates the complementary merits of features by assigning weights to all features. Furthermore, we propose a novel loss function, coined Dynamic Focal Loss, which reshapes the standard cross-entropy loss and dynamically adjusts the weights to correct the distribution of hard samples. We evaluate our method on three datasets, i.e., BSDS500, NYUDv2, and BIPEDv2. CTFN achieves competitive accuracy with fewer parameters and computational costs than state-of-the-art methods. With the exception of the backbone, CTFN only requires 0.1M extra parameters, making the computation 40% less expensive than other state-of-the-art methods. The codes are available at https://github.com/Li-yachuan/CTFN-pytorch-master.