Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Self-Adaptive Gamma Context-Aware SSM-based Model for Metal Defect Detection

Sijin Sun, Ming Deng, Xingrui Yu, Xingyu Xi, Liangbin Zhao

TL;DR

This work tackles metal defect detection under challenging grayscale variations and irregular defect shapes by introducing GCM-DET, a self-adaptive Gamma Context-Aware detector that fuses Dynamic Gamma Correction with a State-Space Model–based multi-scale feature architecture and Focal IoU loss. The Gamma-Carafe module enables adaptive, content-aware upsampling, while the SSM backbone/neck captures long-range spatial dependencies to robustly represent defects across scales. The approach improves detection performance on CD5-DET and generalizes well to NEU-DET and GC10-DET, achieving state-of-the-art or competitive mAP@0.5 with notable reductions in some computational costs compared to larger baselines. This combination of grayscale robustness, multi-scale fusion, and stochastic state-space modeling offers a practical route to reliable industrial defect detection, with future work aimed at efficiency improvements and dataset expansion.

Abstract

Metal defect detection is critical in industrial quality assurance, yet existing methods struggle with grayscale variations and complex defect states, limiting its robustness. To address these challenges, this paper proposes a Self-Adaptive Gamma Context-Aware SSM-based model(GCM-DET). This advanced detection framework integrating a Dynamic Gamma Correction (GC) module to enhance grayscale representation and optimize feature extraction for precise defect reconstruction. A State-Space Search Management (SSM) architecture captures robust multi-scale features, effectively handling defects of varying shapes and scales. Focal Loss is employed to mitigate class imbalance and refine detection accuracy. Additionally, the CD5-DET dataset is introduced, specifically designed for port container maintenance, featuring significant grayscale variations and intricate defect patterns. Experimental results demonstrate that the proposed model achieves substantial improvements, with mAP@0.5 gains of 27.6\%, 6.6\%, and 2.6\% on the CD5-DET, NEU-DET, and GC10-DET datasets.

Self-Adaptive Gamma Context-Aware SSM-based Model for Metal Defect Detection

TL;DR

This work tackles metal defect detection under challenging grayscale variations and irregular defect shapes by introducing GCM-DET, a self-adaptive Gamma Context-Aware detector that fuses Dynamic Gamma Correction with a State-Space Model–based multi-scale feature architecture and Focal IoU loss. The Gamma-Carafe module enables adaptive, content-aware upsampling, while the SSM backbone/neck captures long-range spatial dependencies to robustly represent defects across scales. The approach improves detection performance on CD5-DET and generalizes well to NEU-DET and GC10-DET, achieving state-of-the-art or competitive mAP@0.5 with notable reductions in some computational costs compared to larger baselines. This combination of grayscale robustness, multi-scale fusion, and stochastic state-space modeling offers a practical route to reliable industrial defect detection, with future work aimed at efficiency improvements and dataset expansion.

Abstract

Metal defect detection is critical in industrial quality assurance, yet existing methods struggle with grayscale variations and complex defect states, limiting its robustness. To address these challenges, this paper proposes a Self-Adaptive Gamma Context-Aware SSM-based model(GCM-DET). This advanced detection framework integrating a Dynamic Gamma Correction (GC) module to enhance grayscale representation and optimize feature extraction for precise defect reconstruction. A State-Space Search Management (SSM) architecture captures robust multi-scale features, effectively handling defects of varying shapes and scales. Focal Loss is employed to mitigate class imbalance and refine detection accuracy. Additionally, the CD5-DET dataset is introduced, specifically designed for port container maintenance, featuring significant grayscale variations and intricate defect patterns. Experimental results demonstrate that the proposed model achieves substantial improvements, with mAP@0.5 gains of 27.6\%, 6.6\%, and 2.6\% on the CD5-DET, NEU-DET, and GC10-DET datasets.

Paper Structure

This paper contains 22 sections, 27 equations, 5 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Methodology
  4. Detection Framework
  5. Gamma-Carafe
  6. Dynamic Gamma Correction
  7. Kernel Prediction.
  8. Feature Reassembly.
  9. Channel Compression.
  10. Content Encoder.
  11. Pixel Normalization.
  12. State Space Model
  13. Experiment
  14. Datasets Preparation
  15. Experimental Environment
  16. ...and 7 more sections

Figures (5)

  • Figure 1: Using different Gamma $\gamma$ coefficients can enhance image quality by adjusting exposure and shadows. Multiplying pixel values by $\gamma$ applies a linear scaling transformation. A Gamma coefficient greater than 1 brightens the image, while one less than 1 darkens it. This adjustment improves the visibility of the image in specific situations.
  • Figure 2: Proposed framework includes a Dynamic Gamma Block, where the Gam-Car model applies adaptive Gamma correction to the original image based on variations in $\mu$ and $\sigma$, while the Pix-shuffling provides super-resolution sampling for image scaling. An ODSS Block and a Vision Clue Merge Block are employed to further enhance the image processing.
  • Figure 3: ODSSBlock
  • Figure 4: Precision-recall metrics for the GCM-DET model.
  • Figure 5: The impact of different window sizes on SSMs performance. Proposed model acheves best in each classes.