Why my paper sux

The stuff below is copy-paste from the e-mail of acceptance that I received from ACCV after they reviewed my paper submission.

The following reviews were received for your paper: 

Reviewer: # 1 
Paper summary and major contribution made: 
In this paper, the author(s) are describing about wheelchair detection by
the CV for the wheelchair challenged people. The wheelchair model for the
detection and tacking is made by two parallel circular disks and a head
model (maybe ellipsoid).The head model is located vertically above the
midpoint of the circular disks.Summarize the main technique for wheelchair
detection that have been made in this paper:1. Background subtraction by the
RGB min-max boundary for the wheelchair region segmentation in the image.2.
RGB color based skin region detection for the head part detection.3. Hough
transform for the wheel detection.Through these techniques, the author(s)
realized wheelchair detection and 3D position estimation with the calibrated
camera. The result is not perfect, because the farther wheel has occlusion
by near wheel, so it is difficult to find wheels by the Hough
transformation. The author(s) mentioned about wheelchair tracking by the
optical flow at the introduction, but I could not find this detail.

Detailed comments: 
originarity:It seems technical originality is very few. However the
application which specified wheelchair detection is rare.technical
correctness:At the technical correctness, I could not found any problems.
usefulness significance:To find wheelchair and get it location by the CV it
will be some kind of help for wheelchair challenged people, but I could not
understand the needs from this paper, concretely. So, I can't discuss it
usefulness and significance. suitability for ACCV:The main technique is
based on computer vision, so suitability is enough for ACCV.

Additional comments: (regarding whether the paper describes work in progress) 

Reviewer: # 2 
Paper summary and major contribution made: 
The paper describes a vision based method to find the pose of a wheelchair
in an indoor environment. The position of the wheels is found using an
interesting engineering design for the Hough transform for ellipses (the
camera is deliberately positioned to look down on the wheelchair to remove
ambiguities in determining which circle pose is correct) and a colour
learnign algorithm to find the head. The results show the algorithms working
on a number if images and failure modes are illustrated.

Detailed comments: 
Originality: Originality is only present through the design of the system as
the colour and ellipse algorithms have already been published. However the
design of the overall system is original but it is debatable if it is a
significant advance.Technical correctness: Okay.Usefulness and significance:
This is an interesting scenario but once can't help feeling this is an
artificial situation and perhaps a better system is to use transponders on
wheelchairs for proximity detection etc. The assumptions stated at the
beginning of the paper are somewhat harsh and I believe the system would
break down regularly in a real world situation.Contribution: Mainly on the
overall system design which is not that significant.Suitability to ACCV:
Would be interesting to some people.The paper needs more examples of
wheelchairs (different types with different wheels, and the assumptions need
to be relaxed).

Additional comments: (regarding whether the paper describes work in progress)