n this work, we try to explore and explain content based image retrieval technique for mental illness early detection based on gesture expression. Gesture expression based to recognize mental illness due to gesture has multidimensional and may features for calculation. A technique used to detect and recognize facial expression called Content Based Image Retrieval or CBIR, in this technique needed gesture image training and referencing. This research also proposed to construct an accurate method or algorithm to detect and recognize whether one’s suffers mental illness or not. In this research was carried out using gesture image database and gesture without obstacles (hat, moustache, glasses, etc). Research uses more than 5,000 gesture images with gesture which collected from Lampung mental illness hospital and from the internet. Research produce an image gesture retrieval result quite good in term of precession and recall parameters.