We aim to create n end to end solution for the problem of writing medical reports from medical images. Creating medical reports from images such as XRays and other scans may be error prone and tedious. AGMIR shall allow medical practitioners to automatically generate medical reports and also organize them effectively. We use a multilabel classifier to predict the tag to every part of the image and then an LSTM based sentence model to generate descriptions for the tags. The project borrows it's idea from the research paper On the Automatic generation of medical reports : https://arxiv.org/abs/1711.08195