mohitminhas / multi-object-compressive-tracking Goto Github PK

Readme for Multi Object Compressive Tracking

By:
Mohit Minhas
P. Deepak Prashanth

Mentor:
Dr. Jagdish Raheja

---------------------------------------------------------------------------------------------------------------------------------------------
Procedure
> specify the video path in RuntrackerMulti
> run mexCompile.m to generate mex files 
> run RuntrackerMulti.m
> specify number of objects through command line
> select bounding boxes
The parameters in the main function “Runtracker.m” can be tuned as follows
1.	“trparams.init_postrainrad” is the search radius for the positive sample; This parameter can be set 4~8. If the object moves very fast, a large parameter should be used to contain more positive samples.
2.	“trparams.srchwinsz” is the search radius for the search window at the new frame; This parameter can be set 15~35. If the object moves fast, a larger parameter should be used to contain the object.
3.	“lRate” is the learning rate parameter. This parameter can be set 0.7~0.95; If the object changes fast, a small “lRate” should be used to weight more on the new frames.
4.	“ftrparams.maxNumRect” is the maximum number of nonzero entries at each row of the random matrix. The parameter can be set 4 or 6. If the appearance of the object varies much, 6 should be used to contain more discriminative features.


Reference
---
RealTime
Compressive Tracking, Kaihua Zhang(a), Lei Zhang(a), and MingHsuan
Yang(b),
(a) Depart. of Computing, Hong Kong Polytechnic University, (b) Electrical Engineering and
Computer Science, University of California at Merced
---