A Toolchain for Visual Memory

| 0

Type: UROP
Status: Completed August 2015
Student: Lucas Pennati

Within the EU research project RECALL we are investigating how captured images can serve as “memory cues” to help individuals remember certain past events (e.g., meetings, or an evening with friends). Existing wearable “lifelogging” devices allow for the automated capture of photographs throughout the day, e.g., every 30 seconds. However, in order to use these images as memory cues, we need to have an idea of what’s “on” these pictures. This UROP project will investigate opportunities to automatically classify such images along various dimensions, e.g., the amount of blurriness that they contain, their light levels, but also higher-level contextual information such as the number of people (faces) in the image, or if it contains a cup, etc. The student will develop a modular toolbox using OpenCV that can run an arbitrary set of sensors over a folder of images in order to extract contextual information from these images.

For more information contact: