The state of health (SoH) and efficiency of lithium-ion batteries (LIBs) is critical for the safety and maintenance of energy storage systems and electric vehicle applications. In this paper, different health indicators (HIs) are proposed to accurately estimate the SoH of LIBs, features such as charged capacity, discharged capacity, capacity lost during discharge (CLDD), resistance developed during discharge, average battery temperature charging and discharging are extracted as key HIs of the LIBs. A machine learning (ML) algorithm named back-propagation artificial neural network (BP ANN) is used to map the non-linear relationship of the HIs and Soh due to its fast learning rate and ability to handle large amounts of non-linear data. The testing results on private data on three batteries collected at atmospheric temperature, show that the proposed method can accurately estimate SoH of the LIBs. The proposed method does not require additional downtime making it suitable for online monitoring of SoH.