Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Enhancing Industrial Transfer Learning with Style Filter: Cost Reduction and Defect-Focus

Chen Li, Ruijie Ma, Xiang Qian, Xiaohao Wang, Xinghui Li

TL;DR

Style Filter is evaluated on authentic industrial datasets, highlighting its effectiveness when employed before conventional transfer strategies in the deep learning domain, and underscores the effectiveness of Style Filter in real-world industrial applications.

Abstract

Addressing the challenge of data scarcity in industrial domains, transfer learning emerges as a pivotal paradigm. This work introduces Style Filter, a tailored methodology for industrial contexts. By selectively filtering source domain data before knowledge transfer, Style Filter reduces the quantity of data while maintaining or even enhancing the performance of transfer learning strategy. Offering label-free operation, minimal reliance on prior knowledge, independence from specific models, and re-utilization, Style Filter is evaluated on authentic industrial datasets, highlighting its effectiveness when employed before conventional transfer strategies in the deep learning domain. The results underscore the effectiveness of Style Filter in real-world industrial applications.

Enhancing Industrial Transfer Learning with Style Filter: Cost Reduction and Defect-Focus

TL;DR

Style Filter is evaluated on authentic industrial datasets, highlighting its effectiveness when employed before conventional transfer strategies in the deep learning domain, and underscores the effectiveness of Style Filter in real-world industrial applications.

Abstract

Addressing the challenge of data scarcity in industrial domains, transfer learning emerges as a pivotal paradigm. This work introduces Style Filter, a tailored methodology for industrial contexts. By selectively filtering source domain data before knowledge transfer, Style Filter reduces the quantity of data while maintaining or even enhancing the performance of transfer learning strategy. Offering label-free operation, minimal reliance on prior knowledge, independence from specific models, and re-utilization, Style Filter is evaluated on authentic industrial datasets, highlighting its effectiveness when employed before conventional transfer strategies in the deep learning domain. The results underscore the effectiveness of Style Filter in real-world industrial applications.
Paper Structure (14 sections, 11 figures, 4 tables)

This paper contains 14 sections, 11 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Industrial Surface Defect Detection
  4. Style Transfer
  5. Methodology
  6. Basic Idea
  7. Style Filter
  8. Experiments
  9. Datasets
  10. Implementation Details
  11. Style Filter Results
  12. Validation on Transfer Learning Strategies
  13. Discussion
  14. Conclusion

Figures (11)

  • Figure 1: Basic idea of our Style Filter (SF): style can be employed to describe the differences between samples from various sources. The idea is verified by the visual outcomes of a style transfer experiment conducted on magnetic tile samples collected from various sources. (Best viewed in color.)
  • Figure 2: Overall flowchart of our Style Filter.
  • Figure 3: Specific structure of mapping images to style space. The structure can be primarily divided into two steps: extracting style features and feature fusion.
  • Figure 4: Samples from different manufacturers and the surface defects. These samples manifest numerous discrepancies across several dimensions: actual shape variations of magnetic tiles, variations in backgrounds, and differences in contrast stemming from diverse shooting angles employed to capture distinct magnetic tile surfaces.
  • Figure 5: Methods to determine the optimal value of $k$ for source domain clustering.
  • ...and 6 more figures