Java Software Package For Partial Object Recognition
We have tested our algorithm using  a selected face from random
image in the database of 4000 images.   This selected face is used
as input full or partial face to search in 4000 image database for
exact and relevant faces.  The simulation results have shown that it
always provided correct search for retrieving relevant images.  Our
unique technology is to search correctly for relevant images based
on partial or partial component image where no other technology in
the world can do, based on our knowledge.


Partial Image Search Video Demo
  • Consulting
  • Licensing
Technologies  We Provide
Partial Object/Face Recognition for Internet and smart Phone
Original Image
Partial Input Object (Extracted
from Original Image)
ADC Software
Similar of Partial
Input  Object
Similar of Partial
Input Object
Similar of Partial
Input Object
Statistical Study for Partial and Scaling Object/Face Recognition
The statistical study has been done to demonstrate how robust our algorithm for partial object/face recognition.  Our
algorithm is generic technique for general object as long as the input object contains sufficient features for detection and
recognition.  We used face as object due to the available sufficient face databases and this is only reason.   In this study we
cropped 150 single eye set  and 150 mouth and nose set from 450 images from Caltech database and evaluate with 1970  
images combined Caltech and Switzeland database.  We also studied the random scaling of cropped image served as
partial object/face input to emulate close to real world applications.  The details are provided in the white paper below:

Statistical Study for Partial and random Scaling Object/Face Recognition

The some of eye partial inputs are shown below:
Cropped image as mouth and nose partial input (no scaling)
Random scaling of above image as emulated real world partial face
(scaling numbers: 0.4973   0.8766   0.5867   0.7171   0.4994