The relative density of coarse-grained soils is calculated from the void ratio obtained when the soil is packed in its loosest and densest states in comparison with the maximum void ratio to that of the void ratio in its natural state. The factors that influence the relative density can broadly be grouped under three heads: soil particle-related, compaction-related, and soil moisture related. The factors related to soil particles are particle size distribution, particle shape, and soil structure. In the literature, studies exist to establish the dependency of particle sizes D50 on emax and emin. Furthermore, some of these studies even gave mathematical models to predict emax and emin from the D50 sizes obtained from grain size distribution. The current study is based on the meticulous development of machine learning and deep learning models for poorly graded sands. These models predict the emax and emin based Uniformity Coefficient (Cu) and Coefficient of Curvature (Cc). A vast dataset is meticulously prepared from 180 peer-reviewed journals and conference proceedings, ensuring a comprehensive representation of the field. This dataset includes both emax, emin values and the corresponding either Cu and Cc values or grain size distribution curve for sands so that Cu and Cc are calculated by extracting D10, D30 and D60. The models are developed using Multivariate Regression, Support Vector Regression, Random Forest Regression and Artificial Neural Networks. The model developed using these techniques are able to predicated more than 75% within the percentage error of less than or equal to 10%.