Leveraging Adaptive Equalization for Enhanced Egg Crack Detection in Classification and Object Detection

This study investigates the efficacy of employing the Adaptive Histogram Equalization (AHE) strategy on the classification and object detection of egg cracks. Leveraging Convolutional Neural Network (CNN) algorithms for classification tasks and YOLOv8 for object detection, the research aims to enhance accuracy and reliability. Initial testing demonstrates promising results with the CNN model exhibiting an accuracy of 94.12% before AHE integration, which surges to 96.47% post-integration. Object detection with YOLOv8 shows a significant improvement, with a mean Average Precision at 50% Intersection over Union (mAP50) rising from 80.9% to 84.8% when incorporating AHE. These findings highlight the efficacy of AHE in enhancing both classification and detection processes for egg cracks.