deletions | additions       diff --git a/subsection_Background_of_the_Study__.tex b/subsection_Background_of_the_Study__.tex  index 9ad2a4d..f5d8bdd 100644  --- a/subsection_Background_of_the_Study__.tex  +++ b/subsection_Background_of_the_Study__.tex 
...
  Human body part recognition and tracking has an extensive variety of uses. Before, camera-based movement used frameworks that require unwieldy markers or suits. Lately, research has concentrated on recognition and tracking without camera-based marker frameworks. The intricacy of such frameworks with respect to picture handling depends generally on how a scene is caught. When 2D cameras are utilized, tracking issues, e.g. the assortment of human movements, the impediments between appendages or with other body parts, and the affectability of light changes, are hard to resolve \cite{Gonz_lez_Ortega_2014}. So as to give a more adaptable and vigorous methodology, motion acknowledgment can be seen as a grouping issue, comprising in doling out one mark or class to a motion in a manner that is predictable with the accessible information about a certain issue.\\    For managing an arrangement issue, machine learning systems can be used. These systems utilize a motion preparing set in which every motion is marked to create a classifier \cite{Iba_ez_2014}. However, researchers identify four noteworthy difficulties in vision based human activity acknowledgment. The first is low-level difficulties. Impediments, messed foundation, shadows, and changing brightening conditions can deliver troubles for movement division and modify the way activities are seen. This is a noteworthy trouble of movement acknowledgment from RGB recordings. The presentation of 3D information generally lightens the low-level challenges by providing the structure data of a scene. The second test is perspective changes. The same activities can create an alternate "appearance" from alternate points of view \cite{Aggarwal_2014}. The Kinect sometimes has issues in its calibration as well as some residual errors in close range measurements \cite{smisek20133d}. Lastly, since the Kinect sensors focus on tracking large body segments at a time, in small segments with more complex parts like hands, segmentation is more likely to experience inaccurate recognition \cite{ren2013robust}. This inaccuracy is similar to the way that little mistakes in marker situations and delicate tissue curios are bringing about bigger blunders in the estimation of the joint foci joints  and relative fragment introductions \cite{Bonnech_re_2014}.