Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

High-Performance Fine Defect Detection in Artificial Leather Using Dual Feature Pool Object Detection

Lin Huang, Weisheng Li, Yujuan Tan, Linlin Shen, Jing Yu

TL;DR

YOLOD demonstrated outstanding performance on the artificial leather defect dataset, achieving an impressive increase of 11.7% - 13.5% in AP_50 compared to YOLOv5, along with a significant reduction of 5.2% in the error detection rate.

Abstract

In this study, the structural problems of the YOLOv5 model were analyzed emphatically. Based on the characteristics of fine defects in artificial leather, four innovative structures, namely DFP, IFF, AMP, and EOS, were designed. These advancements led to the proposal of a high-performance artificial leather fine defect detection model named YOLOD. YOLOD demonstrated outstanding performance on the artificial leather defect dataset, achieving an impressive increase of 11.7% - 13.5% in AP_50 compared to YOLOv5, along with a significant reduction of 5.2% - 7.2% in the error detection rate. Moreover, YOLOD also exhibited remarkable performance on the general MS-COCO dataset, with an increase of 0.4% - 2.6% in AP compared to YOLOv5, and a rise of 2.5% - 4.1% in AP_S compared to YOLOv5. These results demonstrate the superiority of YOLOD in both artificial leather defect detection and general object detection tasks, making it a highly efficient and effective model for real-world applications.

High-Performance Fine Defect Detection in Artificial Leather Using Dual Feature Pool Object Detection

TL;DR

YOLOD demonstrated outstanding performance on the artificial leather defect dataset, achieving an impressive increase of 11.7% - 13.5% in AP_50 compared to YOLOv5, along with a significant reduction of 5.2% in the error detection rate.

Abstract

In this study, the structural problems of the YOLOv5 model were analyzed emphatically. Based on the characteristics of fine defects in artificial leather, four innovative structures, namely DFP, IFF, AMP, and EOS, were designed. These advancements led to the proposal of a high-performance artificial leather fine defect detection model named YOLOD. YOLOD demonstrated outstanding performance on the artificial leather defect dataset, achieving an impressive increase of 11.7% - 13.5% in AP_50 compared to YOLOv5, along with a significant reduction of 5.2% - 7.2% in the error detection rate. Moreover, YOLOD also exhibited remarkable performance on the general MS-COCO dataset, with an increase of 0.4% - 2.6% in AP compared to YOLOv5, and a rise of 2.5% - 4.1% in AP_S compared to YOLOv5. These results demonstrate the superiority of YOLOD in both artificial leather defect detection and general object detection tasks, making it a highly efficient and effective model for real-world applications.
Paper Structure (16 sections, 16 equations, 9 figures, 7 tables)

This paper contains 16 sections, 16 equations, 9 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Dataset
  3. Microsoft coco: Common objects in context (MS-COCO)
  4. Artificial leather dataset (ALD)
  5. YOLOD
  6. Dual Feature Pool
  7. Interference Feature Filtering
  8. Adaptive Multi Positives
  9. Eliminate Output Sensitivity
  10. Experiments(ALD)
  11. Experiments(MS-COCO)
  12. Ablation Experiment
  13. Comparative experiment
  14. Comparison with the state-of-the-art
  15. Conclusion
  16. ...and 1 more sections

Figures (9)

  • Figure 1: (a) illustrates the comparison of AP on MS-COCO between the proposed YOLOD and YOLOv5. (b) illustrates the comparison of AP$_S$ on MS-COCO between the proposed YOLOD and YOLOX. (c) illustrates the comparison of AP$_{50}$ on ALD between the proposed YOLOD and YOLOv5. (d) illustrates the comparison of AE on ALD between the proposed YOLOD and YOLOv5.
  • Figure 2: Two types of artificial leather defects are black spots and white spots.
  • Figure 3: Figure (a) presents the external view of the mechanical structure, while Figure (b) shows its perspective view.
  • Figure 4: The FPN+PAN structure. The orange dashed arrow represents the network path of the detection head D3, while the green dashed arrow represents the network path of the detection head D1. It is evident that the orange network is deeper than the green network. Partial convolution and SPPspp are not included from the figure.
  • Figure 5: The diagram illustrates the DFP+PAN structure. The green box contains two CSPcsp structures formed by the combination of three-level features. These two CSP structures constitute two feature pools of different scales and output features to the detection head. Partial convolution and SPP are excluded from the figure.
  • ...and 4 more figures