Freshy Pal: A Proposal of an Interactive Web based Tool to Automatically Detect Microbial Growth on White Bread

Bread has long been a staple food in kitchens across the country and is devoured by billions of people worldwide. Despite being so widely consumed, the bread usually only remains fresh for a short while. Bread can develop molds and microbial growth, making it unhealthy to ingest and potentially harmful to one's health. As a result, it is necessary to determine whether the bread is safe to consume. Hence, Freshy Pal, an interactive web-based application tool was created, that makes significant use of new technologies such as image processing, machine learning, and deep learning to identify bread spoilage given an image of the slice of bread. The primary goal of designing this application is to assist users in determining whether bread is safe to ingest because bread spoilage can cause allergic responses and respiratory difficulties, further removing the clearly spoilt portion of the bread and consuming the remainder is equally harmful. So, after Freshy Pal classifies bread as unsafe to consume, it instructs the users on how to properly dispose of and handle the moldy bread. Freshy Pal also educates its users on how to keep their bread safe from mold growth. Two different methodologies were used to develop the system. The first approach employed extensive Image Processing and Machine Learning methods. The following machine learning algorithms were used: Logistic Regression, Support Vector Machine, Decision Tree, Random Forest, Naive Bayes, and Ada Boost. Random Forest and Ada boost achieved the greatest accuracy of 96.67%. The second approach employed Deep Learning-based Image Classification utilizing Convolutional Neural Networks, here an accuracy of 98% was achieved using the pre-trained model VGG16. The outcomes of both techniques are contrasted on a variety of grounds. The deep learning-based method achieved higher accuracy, but it required longer training time and larger data size.