Download Full Text (6.5 MB)
Karanveer Gill, CIS494: Research in Computer Information Systems
Faculty Mentor(s): Professor Sarbani Banerjee, Computer Information Systems
This project will be studying how to detect people in images who appear miniscule. Small people can be identified as low as 20 pixels. This project has some inspiration from satellite images. It can almost replicate how a satellite image can detect a person. For example, if a picture was taken from a satellite, the identification of small persons will only be beneficial for investigations. This project will be written by using Python language and Jupyter Notebook will be used as the Integrated Development Environment (IDE). Windows Docker will be utilized to access a COCO annotation. This will be accessed within the Python program, referencing a JSON file. A dataset will be downloaded with help from this COCO annotator. This data annotation will be designed to detect tiny persons. The purpose of the project is to immediately detect persons in a picture that are not in a crowded area. Images will mainly be aerial shots where the people in the image will look tiny in size. There will be an indication that the object, in this case the “tiny-person,” is identified. This project will also attempt to zoom-in the photo, almost to replicate a zoomed-in camera or satellite image. With the identification of tiny people, the object detected will turn out to be human. An objective of this research is to mimic a satellite image or camera identifying potential suspects.
Gill, Karanveer, "Are Those Ants?" (2021). Computer Information Systems and Engineering Technology. 15.