DEEP LEARNING MODEL FOR FRUIT QUALITY DETECTION AND EVALUATION

Abstract

 Ensuring the excellence of fruits holds paramount significance in safeguarding human well-being. The automated identification process assumes particular prominence within the realm of food industry and agriculture, offering time-saving benefits and shielding individuals from potential health complications. Fruit quality plays a crucial role in safeguarding human health, making automated detection especially significant in the food industry and agriculture. By leveraging machine learning algorithms, this system offers the potential to save time and mitigate health risks,which is made of CNN(Convolutional Neural Network). The major goal is to make it more accurate with less loss percentage and also predict fruit quality rather than only predicting the fruit. For that we have tried various pre-trained models and found their respective accuracy and loss percentage like Xception, VGG16, Resnet, Inception V3. Fruits used in the project are Apple, Banana, Guava, Lemon, Pomegranate, Orange. For our project, the ResNet model is the most suitable choice. ResNet, short for Residual Network, is a deep convolutional neural network architecture that is known for its effectiveness in image recognition tasks. It has achieved significant success in various computer vision applications.One of the key advantages of the ResNet model is its ability to handle deep networks by utilizing skip connections or residual connections. These connections allow the model to bypass certain layers, enabling the flow of information from earlier layers directly to the later layers. This approach helps alleviate the vanishing gradient problem and enables training of much deeper networks.