Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Color Recognition in Challenging Lighting Environments: CNN Approach

Nizamuddin Maitlo, Nooruddin Noonari, Sajid Ahmed Ghanghro, Sathishkumar Duraisamy, Fayaz Ahmed

TL;DR

This paper tackles the problem of color recognition under challenging lighting conditions. It proposes a CNN-based pipeline that first segments objects with edge detection and then classifies color using a network trained to handle illumination variation. The dataset comprises 250 color-variant images (200 training, 50 testing), and the pipeline employs bounding boxes and color cubes processed by RCCNet for localized color assessment. Experimental results show the CNN-based approach achieves superior robustness compared to traditional methods such as SVM, LSTM, and RF, highlighting its practical value for real-world color detection in diverse lighting environments.

Abstract

Light plays a vital role in vision either human or machine vision, the perceived color is always based on the lighting conditions of the surroundings. Researchers are working to enhance the color detection techniques for the application of computer vision. They have implemented proposed several methods using different color detection approaches but still, there is a gap that can be filled. To address this issue, a color detection method, which is based on a Convolutional Neural Network (CNN), is proposed. Firstly, image segmentation is performed using the edge detection segmentation technique to specify the object and then the segmented object is fed to the Convolutional Neural Network trained to detect the color of an object in different lighting conditions. It is experimentally verified that our method can substantially enhance the robustness of color detection in different lighting conditions, and our method performed better results than existing methods.

Color Recognition in Challenging Lighting Environments: CNN Approach

TL;DR

This paper tackles the problem of color recognition under challenging lighting conditions. It proposes a CNN-based pipeline that first segments objects with edge detection and then classifies color using a network trained to handle illumination variation. The dataset comprises 250 color-variant images (200 training, 50 testing), and the pipeline employs bounding boxes and color cubes processed by RCCNet for localized color assessment. Experimental results show the CNN-based approach achieves superior robustness compared to traditional methods such as SVM, LSTM, and RF, highlighting its practical value for real-world color detection in diverse lighting environments.

Abstract

Light plays a vital role in vision either human or machine vision, the perceived color is always based on the lighting conditions of the surroundings. Researchers are working to enhance the color detection techniques for the application of computer vision. They have implemented proposed several methods using different color detection approaches but still, there is a gap that can be filled. To address this issue, a color detection method, which is based on a Convolutional Neural Network (CNN), is proposed. Firstly, image segmentation is performed using the edge detection segmentation technique to specify the object and then the segmented object is fed to the Convolutional Neural Network trained to detect the color of an object in different lighting conditions. It is experimentally verified that our method can substantially enhance the robustness of color detection in different lighting conditions, and our method performed better results than existing methods.
Paper Structure (15 sections, 5 equations, 9 figures, 1 algorithm)

This paper contains 15 sections, 5 equations, 9 figures, 1 algorithm.

Table of Contents

  1. Introduction
  2. Color Spectrum System
  3. Literature Review
  4. Methodology
  5. Dataset
  6. Experimental Settings
  7. CNN
  8. Pooling Layer
  9. Proposed Framework
  10. Bounding Box Overview
  11. Pusodocode
  12. Experiment and Results
  13. Our Results
  14. Discussion
  15. Conclusion

Figures (9)

  • Figure 1: Spectrum of six different colors, whose brightness changes from dark to light.
  • Figure 2: Spectrum of six different colors, whose brightness changes from dark to light.
  • Figure 3: Spectrum of six different colors, whose brightness changes from dark to light.
  • Figure 4: Spectrum of six different colors, whose brightness changes from dark to light.
  • Figure 5: Training loss and Validation Loss
  • ...and 4 more figures