CSE 454: Advanced Internet Systems
Final Report
Authors:
Jim Li
Jack Hebert
Kiarash Ghadianipour
Abstract
In this paper we present Camoogle++, a location based webcam search engine. Camoogle is designed to allow a user to search for webcams based on their location and provide to the user a graphical search results in the style of Google Image Search. Camoogle accomplishes this through the use of a distributed and highly parallelizable system that could easily be scaled to crawl much larger areas of the internet.
What follows is a high level description of the Camoogle's architecture, the algorithms used to label webcams and localize them, and an analysis of the precision of the localization algorithm. The analysis finds that our algorithm achieves acceptable results for a hand-coded algorithm, with the results strongly relying on the available database of location information.
It is proposed that the next step to improve Camoogle would be to incorporate machine learning based algorithms to help filter out images from being incorrectly labeled as webcams and to localize the true webcams.