Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Selective Feature Connection Mechanism: Concatenating Multi-layer CNN Features with a Feature Selector

Chen Du, Chunheng Wang, Yanna Wang, Cunzhao Shi, Baihua Xiao

TL;DR

This paper proposes a general network architecture to concatenate CNN features of different layers in a simple and effective way, called Selective Feature Connection Mechanism (SFCM), and demonstrates the effectiveness, superiority, and universal applicability of this method on multiple challenging computer vision tasks, including image classification, scene text detection, and image-to-image translation.

Abstract

Different layers of deep convolutional neural networks(CNNs) can encode different-level information. High-layer features always contain more semantic information, and low-layer features contain more detail information. However, low-layer features suffer from the background clutter and semantic ambiguity. During visual recognition, the feature combination of the low-layer and high-level features plays an important role in context modulation. If directly combining the high-layer and low-layer features, the background clutter and semantic ambiguity may be caused due to the introduction of detailed information. In this paper, we propose a general network architecture to concatenate CNN features of different layers in a simple and effective way, called Selective Feature Connection Mechanism (SFCM). Low-level features are selectively linked to high-level features with a feature selector which is generated by high-level features. The proposed connection mechanism can effectively overcome the above-mentioned drawbacks. We demonstrate the effectiveness, superiority, and universal applicability of this method on multiple challenging computer vision tasks, including image classification, scene text detection, and image-to-image translation.

Selective Feature Connection Mechanism: Concatenating Multi-layer CNN Features with a Feature Selector

TL;DR

This paper proposes a general network architecture to concatenate CNN features of different layers in a simple and effective way, called Selective Feature Connection Mechanism (SFCM), and demonstrates the effectiveness, superiority, and universal applicability of this method on multiple challenging computer vision tasks, including image classification, scene text detection, and image-to-image translation.

Abstract

Different layers of deep convolutional neural networks(CNNs) can encode different-level information. High-layer features always contain more semantic information, and low-layer features contain more detail information. However, low-layer features suffer from the background clutter and semantic ambiguity. During visual recognition, the feature combination of the low-layer and high-level features plays an important role in context modulation. If directly combining the high-layer and low-layer features, the background clutter and semantic ambiguity may be caused due to the introduction of detailed information. In this paper, we propose a general network architecture to concatenate CNN features of different layers in a simple and effective way, called Selective Feature Connection Mechanism (SFCM). Low-level features are selectively linked to high-level features with a feature selector which is generated by high-level features. The proposed connection mechanism can effectively overcome the above-mentioned drawbacks. We demonstrate the effectiveness, superiority, and universal applicability of this method on multiple challenging computer vision tasks, including image classification, scene text detection, and image-to-image translation.

Paper Structure

This paper contains 11 sections, 7 equations, 8 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Method
  4. Selective Feature Connection Mechanism
  5. Adding SFCM To Deep ConvNets
  6. Discussion
  7. Experiments
  8. Image classification:
  9. Scene text detection
  10. Image-to-image translation
  11. Conclusion and Future Work

Figures (8)

  • Figure 1: During visual recognition, the feature combination of the low-layer and high-level features plays an important role in context modulation.
  • Figure 2: Different layers of deep convolutional neural networks(CNN) can encode different-level information. High-layer features always contain more semantic information, and low-layer features contain more detail information. In this paper, low-layer features are selected for further use by a feature selector which is generated by high-level features.
  • Figure 3: Low-layer and high-layer features in CNN are combined directly.
  • Figure 4: The proposed SFCM. "$\otimes$" denotes matrix multiplication, "$\oplus$" denotes element-wise sum. (a)Direct connection. (b)Residual connection.
  • Figure 5: The dense block for image classification. (a) A dense block in DenseNet. (b) A dense block with SFCM (shown in orange).
  • ...and 3 more figures