This is an old event-based dataset based on an even older frame-based dataset.
It can be useful for playing with new ideas or teaching, but for anything more serious I suggest looking for the latest datasets on the Event-based Vision Resources page.
The data can be downloaded here (dropbox link).
This directory contains data originally presented in:
Orchard, G.; Meyer, C.; Etienne-Cummings, R.; Posch, C.; Thakor, N.; and Benosman, R., "HFIRST: A Temporal Approach to Object Recognition," Pattern Analysis and Machine Intelligence, IEEE Transactions on vol.37, no.10, pp.2028-2040, Oct. 2015
Available online from:
https://arxiv.org/abs/1508.01176
http://ieeexplore.ieee.org/document/7010933/
There are 3 main files:
"Original.mat": Contains the original raw recording of the characters
"ExtractedMoving.mat": Contains 76 extracted recordings, where each character is tracked, and events not belonging to the character have been removed.
"ExtractedStabilized.mat": Contains the same 76 extracted recordings, but the location (spatial address) for each spike is now relative to the center of the character, so the character appears to stay in place rather than moving across the screen.
This data can be worked with using the AER Matlab Functions from: