Diabetic Retinopathy (DR) is the complexity arises because of raise in glucose level in the blood that causes retinal damage in diabetic patient's eye and leads to vision loss if not early detected and treated. Due to DR, retinal blood vessels get damaged and accumulation of lipoprotein forms Hard Exudates (HE). The early detection of HE in fundus photographs with the use of computer assisted tools helps Ophthalmologists to analyze and prevent vision impairment. Therefore many computer assisted systems were developed using image processing, supervised and unsupervised machine learning and deep learning. In this review, an analysis of latest development of such systems to detect hard exudates is presented. The objective of this paper is to define that systems which uses deep learning algorithms for efficient detection of bright lesions as HEs.