Features and Updates


Hi !! Welcome to the Home of conLSH: Context based Locality Sensitive Hashing



conLSH has been found to save 36.3% of the total time, on an average over the exhaustive search



Action Classification using conLSH produces an average gain of 12.83% over the state of the art methods, in terms of accuracy.



conLSH has been integrated to perform automatic annotation of long and composite videos.






                    ****
Welcome !!

.
An Improved Video Analysis using Context based Extension of LSH

	   Context based Locality Sensitive Hashing (conLSH) finds 
the approximate nearest video clips from a query video, after 
inspecting the contextual similarity present in a video sequence. 
In conLSH, a video frame is hashed in accordance with its left 
and right context frames. Here, the hash value of the ith frame 
in the video sequence is computed based on the values of the 
(i-1), i and (i+1)th frames. The videos having similar contexts 
are hashed together. When the query comes, it is hashed using 
the same conLSH function. The videos that are in the same 
slot with the query, are searched for nearest match.
Designed And Maintained By Angana Chakraborty, Assistant Professor WBES
SOURCE CODE            RESULT            CONTACT ME               HOME
conLSH
Context based Locality Sensitive Hashing