Student research opportunities

Spatio-Temporal Change Detection in Giga-Pixel Images

Project Code: CECS_1177

This project is available at the following levels:
CS single semester, Engn R&D, Honours, Summer Scholar, Masters

Keywords:

Giga-pixel images, spatio-temporal data mining, change detection

Supervisor:

Dr Anoop Cherian

Outline:

Gigavision is a hardware/software system that enables recording and analysis of decades-long time‑lapse images of the environment at a scale of billions of pixels. Gigapixel resolution time-lapse images make possible visual and numeric analysis of phenological and environmental change across a wide spatial range from the level of the individual to the ecosystem. Gigavision data can be embedded in geospatial visualization tools such as Google Earth and Google maps and used in interactive kiosk displays in National Parks and conservation areas. In addition to scientific applications, gigapixel time-lapse datasets can be used to create compelling video content to promote conservation and educate the public about climate change and other slow but significant environmental change. Gigapixel image sets can also be integrated into online interactive educational products.

The project will be co-supervised with Dr. Tim Brown at the Borovita Lab.

Goals of this project

The goal of this project is to build automated tools for processing large scale high-dimensional spatio-temporal data generated by the Gigavision system applied to monitoring large plant eco-systems in the scale of an entire arboretum. In this project, the student will gain exposure to
1. developing software platforms that can scale to handle large datasets,

2. develop computer vision algorithms for image analysis

3. develop machine learning techniques for change detection and pattern mining tasks.

Requirements/Prerequisites

Strong programming skills in C++, Matlab and/or Python are required. A background in computer vision and machine learning is desirable. Also, some web programming experience will be a plus.

Student Gain

Developing cutting technologies for computer vision and machine learning to handle very large scale data.

Links

GigaVision System
Time Science

Contact: