First, we need to find a representation of the data which results in good performance on detecting the plasmodium, and then to have a general representation of the shapes present in the images containing blood smear excluding objects such as leukocyte or the different hemoparasites, so that in future the same platform can be used to identify the other related problems.