A comparison of the complexity between observational KiDS LRGs data provided by Li et al. (2020) and the LRG samples within our non-lensed population from the training data set. It highlights the adjustments made to the LRG population, achieved through strategic augmentation of LRG samples. These modifications aim to enhance the visibility of sources with less representative samples, particularly those with complexity values exceeding 100. By augmenting these samples, the CNN is exposed more to the complex LRGs and, thereby enhancing its ability in identifying and analysing LRGs as non-lensed samples.