Paper Title
Classification of all-Purpose Sand: A Deep Learning Approach

Abstract
Construction sand classification is an essential process in the construction industry. The proper classification of sand plays a crucial role in determining the quality and suitability of sand for specific construction purposes. In general, sand is classified based on its particle size, shape, texture, and colour. Understanding these characteristics can help builders and contractors choose the right sand for their projects.The particle size of sand can also be classified using the Unified Soil Classification System (USCS), which categorizes sand as poorly graded, well-graded, and uniformly graded. In the building sector, differentiating sand based on its texture would be very beneficial as it will save builders, contractors, and construction workers a lot of time and work and speed up their process, resulting in higher efficiency and better results. One such model, which is actually a combination of three different pre-trained deep learning models, is proposed in this study combining machine learning and deep learning approaches. Based on the collection of images that are given to it, the model assists in classifying the sand. The results demonstrate that the first model used, Densenet-169, provided approximately 92% of accuracy, while the other two models, Inception V3 and Xception, provided approximately 85-90% accuracy. The weighted average ensemble model provided the best accuracy of 98%, which was actually the best we got.