this is for holding javascript data
Paul St-Aubin edited Methodology Video Data.tex
over 9 years ago
Commit id: 893fa9eb82b8c5afc6038297daa1c5da4b73d867
deletions | additions
diff --git a/Methodology Video Data.tex b/Methodology Video Data.tex
index 5aabaef..1ada6b3 100644
--- a/Methodology Video Data.tex
+++ b/Methodology Video Data.tex
...
\begin{itemize}
\item \textbf{Parallax error} is mitigated by maximising the subtending angle between the camera and the height of tracked objects. In practical terms this requires a high angle of view or ideally a bird's eye view, tracking objects with a small height to base ratio. Passenger cars are generally more forgiving in this respect than heavy vehicles or pedestrians.
\item \textbf{Pixel resolution} determines measurement precision. Objects further away from the camera experience lower tracking precision than objects near the camera. Error due to pixel resolution is mitigated by placing the camera as close to the study area as possible and using high-resolution cameras, though increases in resolution offer diminishing returns of tracking accuracy.
\item Finally, \textbf{tracking errors} may occur with scene visibility issues or due to limits with current computer vision techniques. These erroneous observations have to be rejected or reviewed manually.
Some attempts Attempts have been made
in recent literature and for this work at validating and optimising tracking accuracy using
search algorithms and the Measures Of Tracking Accuracy (MOTA)
methodology \cite{ettehadieh15systematic}. MOTA is a measure of accuracy that combines multiple sources of error simultaneously at a great level of detail, including false positive and false negative detection as
measures well as spatio-temporal accuracy of
performance \cite{ettehadieh15systematic}. This method measurements along entire trajectories in camera space. The tracking optimisation is replicated for this study using a genetic algorithm that searches heuristically for optimal tracking parameters using MOTA analysis as a measure of fitness from manually annotated video data using Urban Tracker \cite{Jodoin_2014}. See section \ref{tracking_calibration} for
details and results.
\end{itemize}
Depending on the steps taken to minimize tracking errors, feature-based tracking functions best over study areas of up to 50-100~m in length with high-to-medium speed, low-to-medium density flows.