Indoor localization has a significant contribution to various location-based services, mostly provided at different public indoor areas like hospital, office, shopping malls, etc. Since indoor areas are deprived of satellite signals, localization is done based on indoor available sensors like WiFi and Bluetooth. To make an indoor localization system precise and professional, the ground truth data must satisfy the quality and quantity requirements. However, collecting enough data from all location points of a public place is infeasible, as some locations can be inaccessible sometimes. As a result, class imbalance has become a common challenge for indoor localization. In this work, this research challenge has been addressed using conditional generative adversarial network (CGAN)., The entire workflow is presented by an algorithm—Data Augmentations for Minority Location Classes (DAMLoc). The proposed approach is experimented on the benchmark dataset JUIndoorLoc, investigating class imbalance in single-floor and multi-floor scenario. Localization accuracy is obtained using three different machine learning classifiers. Around 8% and 5% improvement in accuracy is observed addressing the single-floor and multi-floor class imbalance, respectively.