Deep Learning Based Anchor Model Used to Classify the Fruit by Using the Digital Image

One of the cornerstones of processer vision is visual object recognition based on deep learning. Bounding boxes are also detected by these models, in addition to locating and classifying many items in an image. Ripeness and over ripeness classification of fruits using digital photographs is the main subject of this paper's study. Using visual clues derived from the photographs, our recommended algorithm examines the fruit's peel to estimate the fruit's categorization. Two “anchor-free” “models, YOLOv8 and Center Net, should be trained using our datasets so that they can make reliable predictions. For feature map up sampling, the Cantered network mostly use ResNet-50 and the deconvolution module De Conv. Predictions for the heatmap are made using the last three layers of convolutional neural networks. A combination of C2f and Csp components allows for efficient and light distribution in the YOLOv8 model. Our evaluation and comparison of the two representations revealed that the C2f element of the YOLOv8 model achieves an exceptional 99.5% accuracy rate, significantly enhancing classification results.