Paul St-Aubin deleted Experimental Results Data Sizes.tex  almost 10 years ago

Commit id: fdb21e15979e3641ba4048fae53cf96e125a0a61

deletions | additions      

         

\subsection{Data Sizes}  In total, 600 hours of video data were collected, 50~\% of which has been fully processed at this time. This represents driving data for over 50,000 unique drivers.  At 30 frames per second, a data collection at an intersection over a period of 12 hours (e.g. 7 AM to 7 PM), over a driving distance of 50 metres and at an average driving speed of 30 km/h, and with an average hourly flowrate of 500 veh/h yields approximatly 90,000 instantaneous feature-grouped measurements per hour. Additionally, each of these observations can have anywhere between 3 to 100 feature tracks associated with. The recommended number of features to aim for is roughly 15-20 per object over time: this yields manageable data sizes (roughly 500 MB of storage per hour of video) while maintinging and adequate level of data richness and object representation. Video storage needs will vary greatly by camera choice, resolution, framerate, and video encoding settings.  Figure~\ref{fig:interaction-sizes} shows the orders of magnitude of the interactions observed in the various sites. Traffic volume and lane arrangement appear to be highly correalted with number of user pairs. This is consistent with the approximation made in Equation~\ref{eqn:data-size}. User pairs per hour should be linearly correlated with interaction-instants. If they are not, it is possible that some site analysis areas are not comparable.         

figures/motion-prediction2/motion-prediction2.png  figures/Speed profiles (1lane) fr/Speed profiles (1lane) fr.png  figures/histo collage/histo collage.png  Experimental Results Data Sizes.tex  figures/interaction data/interaction data.png  Experimental Results Processing Times.tex  methodology.tex